Space and Terrestrial Biotechnology [Reprint 2021 ed.] 9783112539521, 9783112539514

Citation preview

Space and Terrestrial Biotechnology A. Cogoli, A. Tschopp

Biotechnology in Space Laboratories J. E. Zajic, I. I. Inculet, P. Martin

Basic Concepts in Microbial Aerosols K. Schiigerl

Characterization and Performance of Single- and Multistage Tower Reactors with Outer Loop for Cell Mass Production

AKADEMIE-VERLAG BERLIN

Space and Terrestrial Biotechnology

Space and Terrestrial Biotechnology Managing Editor : A. Fiechter

with 136 Figures and 29 Tables

Akademie-Verlag • Berlin 1982

Die Originalausgabe erscheint im Springer-Verlag Berlin—Heidelberg—New York als Volume 22 der Schriftenreihe Advances in Biochemical Engineering

Vertrieb ausschließlich für die D D R und die sozialistischen Länder Alle Rechte vorbehalten © Springer-Verlag Berlin, Heidelberg 1982 Erschienen im Akademie-Verlag, DDR-1086 Berlin, Leipziger Straße 3—4 Lizenznummer: 202 • 100/544/82 Gesamtherstellung: VEB Druckerei „Thomas Müntzer", 5820 Bad Langensalza Umschlaggestaltung: Karl Salzbrunn Bestellnummer: 763 153 9 (6725) • LSV 1315 Printed in G D R DDR 9 6 , - M

Managing Editor Professor Dr. A. Fiechter Eidgenössische Technische Hochschule, Hönggerberg, CH-8093 Zürich

Editorial Board Prof. Dr. S. Aiba

Department of Fermentation Technology, Faculty of Engineering, Osaka University, Yamada-Kami, SuitaShi, Osaka 565, Japan

Prof. Dr. B. Atkinson

University of Manchester, Dept. Chemical Engineering, Manchester/England

Prof. Dr. J. Boing

Röhm GmbH, Chem. Fabrik, Postf. 4166, D-6100 Darmstadt

Prof. Dr. E. Bylinkina

Head of Technology Dept., National Institute of Antibiotika. 3a Nagatinska Str., Moscow M-105/USSR

Prof. Dr. H. Dellweg

Techn. Universität Berlin, Lehrstuhl für Biotechnologie, Seestraße 13, D-1000 Berlin 65

Prof. Dr. A. L. Demain

Massachusetts Institute of Technology, Dept. of Nutrition & Food Sc., Room 56-125, Cambridge, Mass. 02139/USA

Prof. Dr. R. Firn

School of Chemical Engineering, Olin Hall, Ithaca, NY 14853/USA

Prof. S. Fukui

Dept. of Industrial Chemistry, Faculty of Engineering, Sakyo-Ku, Kyoto 606, Japan

Prof. Dr. K. Kieslich

Wissenschaftl. Direktor, Ges. für Biotechnolog. Forschung mbH, Mascheroder Weg 1, D-3300 Braunschweig

Prof. Dr. R. M. Lafferty

Techn. Hochschule Graz, Institut für Biochem. Technol., Schlögelgasse 9, A-8010 Graz

Prof. Dr. K. Mosbach

Biochemical Div., Chemical Center, University of Lund, S-22007 Lund/Sweden

Prof. Dr. H. J. Rehm

Westf. Wilhelms Universität, Institut für Mikrobiologie, Tibusstraße 7—15, D-4400 Münster

Prof. Dr. P. L. Rogers

School of Biological Technology, The University of New South Wales. PO Box 1, Kensington, New South Wales, Australia 2033

Prof. Dr. H. Sahm

Institut für Biotechnologie, Kernforschungsanlage Jülich, D-5170 Jülich

Prof. Dr. K. Schügerl

Institut für Technische Chemie, Universität Hannover, Callinstraße 3, D-3000 Hannover

Prof. Dr. H. Suomalainen

Director, The Finnish State Alcohol Monopoly, Alko, P.O.B. 350, 00101 Helsinki 10/Finland

Prof. Dr. G. T. Tsao

Director, Lab. of Renewable Resources Eng., A.A. Potter Eng. Center, Purdue University, West Lafayette, IN 47907/USA

Table of Contents

Biotechnology in Space Laboratories A. Cogoli, A. Tschopp

1

Basic Concepts in Microbial Aerosols J. E. Zajic, I. I. Inculet, P. Martin

51

Characterization and Performance of Single- and Multistage Tower Reactors with Outer Loop for Cell Mass Production K. Schugerl

93

Biotechnology in Space Laboratories Augusto Cogoli, Alex Tschopp Laboratorium für Biochemie, ETH-Zentrum, CH-8092 Zürich, Switzerland

1 Introduction 2 Biology in Space 2.1 Man in Space 2.2 Cell Biology 2.3 Radiobiology 2.4 Exobiology 3 Biological Payloads on US Missions 3.1 Biostack Experiments 3.2 Tissue Culture Incubator on Skylab 3.3 Separation of Cells by Electrophoresis 3.3.1 Static Discontinuous Electrophoresis 3.3.2 Free-flow Continuous Electrophoresis 3.4 The Viking Mission to Mars 4 Biological Payloads on Soviet Missions 5 The Spacelab 5.1 The First Spacelab Mission 5.1.1 Lymphocyte Cell Cultures on Spacelab 5.2 The Follow-on Spacelab Program 5.2.1 Equipment Provided by NASA 5.2.2 The ESA Biorack 5.2.3 Other Biological Equipment for the Spacelab 6 Bioprocessing of Mammalian Cells in Space 7 Designing Equipment for the Spacelab 8 High-g and Low-g Ground Simulations 9 Outlook 10 Acknowledgements 11 Abbreviations and Acronyms 12 References

2 3 3 4 5 5 10 10 11 13 14 16 18 19 25 26 30 32 32 34 34 35 38 43 45 47 47 48

The advent of the Space Shuttle and of the Spacelab will open new perspectives to biotechnology in space. The objectives of this review are: a) to present an overview on the technological and scientific aspects of biological experiments performed on the past US and Soviet space missions, b) to describe the facilities offered by Spacelab in the future, c) to give practical information on the requirements of flight hardware and on the limits in weight, energy and crew-time. Experiments on Apollo and on the Apollo-Soyuz Test Project have shown that the weightless environment offers advantageous conditions to the processing of mammalian cells. Soviet investigations demonstrated that artificial gravity attenuates some of the disturbing effects of spaceflight observed on animals and plants. An extensive program of preliminary studies should precede large-scale biotechnological applications: Suitable hardware has to be developed in collaboration with ESA and NASA, biological objects should be selected as potential candidates for bioprocessing in space. The preparation of investigations in space should be accompanied by ground high-g simulations in centrifuges and by compensating gravity in clinostats. The exploitation of space resources and the establishment of space colonies is becoming a realistic goal for the next decades.

2

A. Cogoli, A. Tschopp

1 Introduction Over twenty years of manned spaceflights have shown that man can survive and work in space for a prolonged period of time (more than 6 months). The main objectives of the past missions were to show that man can adapt himself to the new environment and to explore the solar system. Astronomy and astrophysics were the disciplines which most profited from the scientific space program. In comparison, facilities for life sciences were less sophisticated and were flown on a limited number of missions. With the advent of the Space Shuttle and of the Spacelab we are at the beginning of a new era in space technology: Time has come for the exploitation of space resources. It is known that weightlessness provides favorable conditions for the production of high-quality materials like metal alloys and glasses. Similarly, preliminary experiments indicate that purification of specialized human and animal cells is sometimes easier to perform in space than in ground laboratories. Therefore, material sciences and bioprocessing are becoming new important disciplines in space sciences. An extensive program of research in space biology will start as soon as Spacelab becomes operational. A basic study on the survival and adaptation of living systems to the space environment has to be performed in parallel with applied research. Every biological experiment in space is a technological challenge since flight hardware must comply with the strict safety and engineering specifications set by the space agencies on the one hand and with the biological requirements of living systems on the other. The objective of this review are: — To give a summary of the results and describe the equipment of experiments in space biology performed on past missions. — To discuss the technological aspects of the preparation and execution of life science projects in space. In particular, we describe our own experiments to be performed on Spacelab; — to draw the attention of the reader to data and observations often not easily available; — to stimulate the scientific community to participate in the activities in space taking profit of the facilities offered by the Spacelab and other space stations; — to analyze the potential benefits of space research; — to describe systems for high- and low-g ground simulations, and our observations with animal cells under altered g conditions. Here we discuss mainly experiments and technology dedicated to the study of the behavior in space of microorganisms and animal cells cultured in vitro. We will not discuss equipment for biomedical research, nor for growing plants in space. However, in Section 2 a we will give a survey of the major achievements of life science experiments in space. Valuable results were obtained on automatic satellites. Unfortunately, complicated mechanisms are often subject to failures. Only the constant presence of man as an intelligent observer and operator can guarantee a critical evaluation of the experiment and a proper function of the equipment. Finally, experiments in space should be, whenever possible, preceded by ground simulation at high and low-g. We will describe a number of interesting results obtained by this approach.

Biotechnology in Space Laboratories

3

2 Biology in Space This section presents an outline of the topics belonging to the field of space biology and of their scientific background. The past and the future of space biology have been outlined in a brochure edited recently by Bjursted ^ Although we will not discuss the technology of growing plants in space laboratories, it must be pointed out that plants are ideal objects for the study of gravitational effects since plant growth is regulated by well identified gravity receptor cells (statocytes) carrying gravitysensitive organelles (statoliths).

2.1 Man in Space The primary question at the beginning of the astronautic era was whether humans can survive at all for a prolonged period of time in space. This question was associated with the development of a life-supporting system (energy, atmosphere and waste disposal) capable of providing adequate living conditions to the crewmen. We can regard this phase as completed although technology will progress toward better and better solutions (see Sect. 9). The physiological problems of man were identified already at the early times of astronautic and were sharply focused after the recent long-duration Soviet flights on the Salyut-6 station. A comprehensive discussion of adaptation of man in space is given in Ref. 2) . A brief summary is presented here. a) The space motion sickness involves mainly the vestibular apparatus. Its occurrence is unpredictable and cannot be detected by ground simulations. It may be accompanied by serious symptoms of indisposition which can prevent astronauts from performing flight operations. b) Cardiovascular changes: within few hours in weightlessness, approximately 2000 ml of body fluids (plasma and interstitial liquids) are shifted from the lower to the upper parts of the body. This effect does not cause serious inconveniences. c) Degradation of bone material and muscle atrophy. These symptoms seem to be in part irreversible. Intense fitness training inflight does not prevent completely the diseases. d) Hematological and immunological changes: A relevant loss of red blood cells, hemoglobin and plasma volume are observed after every mission. Lymphocyte reactivity is often reduced after spaceflight. We will discuss here in more detail only those aspects of the adaptation of man in space which can be investigated by biotechnological applications such as in vitro simulations of cellular events. Interesting examples are the immunological and hematological systems: Reduction of red blood cell mass (2—21 %) and of hemoglobin mass (12—33%) is generally observed after the US and Soviet space mission. The changes are accompanied by a loss of plasma volume (4—16%) 3 _ 9 ) . Erythrocyte and hemoglobin concentrations in the blood remain constant, suggesting that the changes are driven by a feed-back mechanism. Immunological changes consist mainly of reduced T-lymphocyte reactivity. The results of the 96-day and 140-day Salyut-6 missions suggest that the adaptation of the immune system to spaceflight occurs in two stages: The first takes

4

A. Cogoli, A. Tschopp

place during the first 2—3 months in space, the second follows and consists of further weakening of the immune response 9 _ 1 4 ) . It is important to point out that the RBC mass reduction and the depression of lymphocyte reactivity never harmed the health of the crews. The changes reflect physiological adaptation reactions rather than pathological conditions. However, with the advent of the Space Shuttle and of large space stations, the opportunity of flying in space will be offered to a broader community. The selection of crews will be less severe than in the past and therefore the hazard of anemic diseases and infections will be higher. The causes of the changes are not fully understood and further investigations are needed to explain these phenomena. In addition to the medical examinations routinely performed, it will be useful to investigate systematically in vitro some of the important cellular processes which appear to be influenced by the space environment. This kind of experiments, when performed on space laboratories, will permit to discriminate between the effect of stress of spaceflight on the organism and the effect of 0 x g on the biological system under investigation. This will deliver an important contribution to applied and basic research in space. As shown in Sect. 5.1.1 certain aspects of the immune system can be investigated in vitro.

2.2 Cell Biology One of the most appealing features of experiments with living cells in weightlessness or at high-g is the transformation of gravity, a physical entity always constant in our ground laboratories, into a variable parameter like temperature or concentration. Consequently, living organisms which underwent evolution and development in a constant gravitational environment are suddenly confronted with a new situation. Therefore, the survival and proliferation of mammalian cells in altered gravitational fields is a challenging aspect of space biology. The effect of varying the gravitational fields (mainly high-g) has been investigated on a variety of living organisms as long ago as 1806 (see Ref. 1 5 ) for a summary). The studies included plants, frog and sea urchin eggs, bacteria and amoeba, as well as complex organisms such as rat and man. Generally, the effects are more dramatic with increasing complexity of the investigated organism. A more detailed description of experiments with isolated cells in space is given in Sect. 3. Calculations made by Pollard 16) show that the distribution of cell organelles like mitochondria, nucleus, nucleolus and ribosomes may be influenced by gravity provided there is sufficient freedom of movement within the cell. This condition is satisfied in cells larger than 1 |im. Therefore, animal cells should be more subjected to gravitational effects than bacteria. In fact, when bacteria were grown in a centrifuge at 50.000 x g 1 6 ) no effect was observed. However, the calculations of Pollard do not take into account cytoplasmic interactions like those involving cytoskeleton. In fact, intracellular movements are severely impaired by rather rigid structures. Gravity may interfere with cytoplasmic streaming as calculated by Kessler 17) . Folkman and Moscona 18) described the correlation between cell shape and growth: Cells of various lines in suspension are spherical whereas cells adhering to the walls of a culture flask are rather flat. It was found that when cells are converted from a flat shape into a spheroidal shape, cells incorporate less 3 H-thymidine into DNA. These findings provide further

Biotechnology in Space Laboratories

5

arguments in favor of experiments studying the effect of gravity on cell proliferation since one can expect that cell shape is influenced somehow by forces and tensions related to gravity. However, Todd 19) detected no influence of gravity on the behaviour of the mitotic spindle in cultured human kidney cells. No effect of spaceflight was observed on human embryonic cells flown on Skylab 3 15) . However, none of the experiments mentioned involved cell differentiation events. Our findings with lymphocytes cultured at different g-levels are described in Sect. 8. Briefly, high gravity has a stimulatory, simulated low gravity an inhibitory effect on lymphocyte activation by mitogens.

2.3 Radiobiology Several kinds of radiations were detected in space ranging from UV and X-rays to high-energy particles. Among the high-energy particles (electrons, protons and heavy ions) the most important for radiobiologists are the HZE (high-charge and highenergy) particles, mainly iron nuclei. Although astronauts were hit by HZE particles, registered as flashes of light, during the past missions, no consequences for their health has been reported. Calculations for a standard Space Shuttle/Spacelab orbit show that flux of Fe nuclei is approx. 2.5 Fe c m - 2 sr per day at solar maximum activity and 0.8 Fe c m - 2 sr per day at solar minimum activity. About 10—20% of Fe ions have energies less than 500 MeV/nucleon, and are highly ionizing. About 25 % of Fe nuclei will interact with an aluminium wall of 5 g c m - 2 thickness. This is the minimum wall thickness encountered by a particle penetrating into the Spacelab. Particular caution should be taken in case of solar flares: the radiation consists of 95 % protons. Surface doses may be higher than 1000 rads; however, the penetration of the radiation is quite low. The effect of cosmic radiation on biological objects was studied on Biosatellite II, on Apollo 16/17 and on the ASTP missions (see Sect. 3.1). Cell damage like cell death, tumor induction and genetic mutations may occur at different levels, depending on the organelles hit. Within certain limits, the study of cosmic radiation effects can also be performed on the ground with the accelerators now available. However, a combined effect of radiation and microgravity can be achieved only in space laboratories. Finally, every biological experiment in space should take into account the effect of radiation. Dosimeters will record radiation in different locations on Spacelab.

2.4 Exobiology The objectives of this discipline are to study origin of life in the universe and the detection of extraterrestrial life. Simulation experiments with primordial elements like hydrogen, oxygen and nitrogen showed that simple molecules (methane, water, ammonia) can be formed in a primitive atmosphere. These molecules can react with one another and produce amino acids, purines, pyrimidines and carbohydrates, which are the essential constituents of the molecules of life. The first and only attempt to search for extraterrestrial life through a biological approach was performed on the Viking missions in March 1975 (see Sect. 3.4). It may be of interest to the reader to know that the total number of technological civilisations which have appeared over the entire history of our galaxy has been estimated to be around a billion.

6

A. Cogoli, A. Tschopp -o c c o

-C Cs Û0J3 •o G S S V< uu O oÖ oa «aÖC ato co. C M OOO OO O ü ifi Ííë O ^ O 2 I * OH .2 o S 00 u V. & O 3 ca E JU

u

00

o Bo « ta O.

1 a. V Oh

6 f

•s

% a* ü 0 1 Vi a> a> u. «

c

60 S

B

ë OH

j «y ed

u o* u

o

co

e'"

es

Biotechnology in Space Laboratories

•o e

c .5

13 6 CL o ° 3 •o Q

o. '3 O" o •C a, wi BO to X)

o 0L,

«

B a i >- a ^ • p

cd s

10

A. Cogoli, A. Tschopp

3 Biological Payloads on US Missions In this section experiments performed aboard manned and unmanned US space stations are reviewed. As pointed out before, very little research has been done aboard manned satellites. A number of interesting experiments were carried out on automated satellites, mainly on the US Biosatellite II. However, some projects failed for mechanical reasons. We discuss here in detail those projects which are most relevant to space biotechnology. A concise summary of all biological experiments carried out on US missions has been recently published by NASA 20) . A consistent program of basic and applied research in life science will start when the Spacelab becomes operational. There, facilities like incubators, centrifuges, microscopes, freezers and refrigerators will offer to the user acceptable working conditions (see Sect. 5). In addition, a selected group of scientists is presently trained by NASA as mission and payload specialists for biological experiments. However, we must always keep in mind that safety requirements and constraints like weight, energy and crew time will always limit the scientific goals of the investigations. Two aspects are relevant: a) The instrumentation carried by the space vessel, and b) the effects of the space environment on growth and behaviour of the organism studied. Table 1 compiles the biological experiments carried on US missions so far 20) . Biomedical investigations on humans or experiments with animals or plants are not discussed. The instrumentation employed was, except for few cases, very simple and mainly consisted of passive containers (petri dishes or sealed ampoules) in which the microorganisms were kept in a nutrient medium. The temperature control was rather primitive and mainly consisted of thermos containers. Man power invested in biological experiments was very limited or nil. The duration of the missions was in most instances too short for an extensive study of the effect of spaceflight on growth and survival of microorganisms. The interest of the investigators was directed toward the detection of radiation damages rather than toward the effect of weightlessness per se. Two of the most important missions, namely Biosatellite I and II either failed or landed too early, upsetting the schedule of most experiments. Therefore, it would be premature to draw conclusions on the adaptation of living systems in space from the studies which have hitherto been performed. We should use the experiences gained for a better choice of the system to investigate and for improving the reliability of the equipment. More biochemical parameters should be analyzed in the future. D N A , R N A and protein biosynthesis are useful indicators of intracellular changes still unexplored in weightlessness. Table 1 shows that very little has been done in this direction particulary with animal cells. In addition, differentiation processes have never been studied in vitro. Lymphocytes and hemopoietic cells are good test objects for this purpose (Sect. 5.1.1). Four relevant projects performed on US missions are discussed here in more detail.

3.1 Biostack Experiments Biostack was one of the first sophisticated experimental devices carried out in space. It consisted of a passive container which did not require power or crew interface. Biostacks I, II and III were flown on the Apollo 16,17 and Apollo-Soyuz Test Project

Fig. 1. Schematic representation of the biostack conception. Left: Monolayers of biological objects are sandwiched alternately with nuclear track detector sheets. Right: Biostack flight container. CN: cellulose nitrate, PC: polycarbonate, PVA: poly-(vinyl alcohol) (courtesy of H. Biicker, DFVLR, Frankfurt)

(ASTP) missions, respectively. The biostack experiments were designed and evaluated by a group of 30—40 investigators in Europe and USA 20) . The objective of these experiments was to study the effect of cosmic radiation on a variety of biological objects (Table 1): Microorganisms, eggs and plants seeds were embedded in poly(vinyl alcohols) between photographic emulsion layers (Fig. 1). This concept first proposed by Eugster 21) and developed by Biicker 2z) allows the identification of the object hit by the radiation and to track the path of the penetrating particle through the biological object. Eggs and plant seeds were grown after landing in the ground laboratories. Several effects of radiation were detected. Most interesting are Artemia salina eggs developed to individuums with abnormalities in the extremities, in the abdomen and in the thorax. Zea mays seeds produced plants with leaves showing large unusual yellow strips. Biostack is a typical example of a simple, passive equipment without energy or crew operation requirements with a high scientific yield. However, the equipment is suitable only for investigations of the effect of radiation whereas gravitational effects cannot be detected with this approach.

3.2 Tissue Culture Incubator on Skylab The effect of spaceflight on human embryonic lung cells WI-38 was studied during the 56-day Skylab 3 mission by Montgomery et al. 1 5 , 2 3 ) . The experiment was

12

A. Cogoli, A. Tschopp

performed with one of the most sophisticated equipment ever used in space biology24'. It consisted of a fully automated tissue culture package called Woodlawn Wanderer 9, designed to achieve four major objectives: — To keep cells alive in culture by supplying fresh medium at 36 °C for several days; — to record phase contrasted pictures with a time-lapse camera for 28 days; — to fix several specimens of cells at given times; — to recover living cells after flight for further culturing in the ground laboratory. The instrument weighed 10 kg, measured 4 0 x l 9 x 17 cm and was powered by 28 DC with an average consumption of 16 W at 10 °C room temperature. The package was separated into two main compartments: A camera-microscope section and a redundantly sealed growth-curve experiment section (Fig. 2). The camera-microscope section consisted of two independent 20 and 40-power camera microscopes. The pictures were recorded on a 16 mm microfilm. The cells were grown on glass in perfusion chambers filled with cultures containing 7000 cells/ml. After the cells became attached to the lower glass disk, the chamber was fixed in the microscope and focused. Each chamber had a volume of 105 jal and was connected to an automatic perfusion system adding fresh medium every 12 hours. The growth-curve experiment section was contained in a module easily removable for biological servicing and consisted of two separated identical and independent

13

Biotechnology in Space Laboratories Media-pump motor

A

Circuit board

Fixing valve Fixative-pump motor

T~eT) n

Media pump-

>-

A

— Media preheater Electrical connector

Fig. 3. Growth-curve experiment section of Woodlawn Wanderer incubator (courtesy of R. G. Thirolf)

units (Fig. 3). Each unit was composed of nine cell culture chambers. Here too cells were fed automatically. At pre-programmed times, a fixative (5 % glutaraldehyde in Earle's balanced salt solution) was added. On mission day 12, four of the nine chambers were not fixed but maintained at room temperature for the rest of the mission. The cells were recovered alive and further cultured in the ground laboratory. An extensive analysis of the collected data showed that the space environment did not produce detectable effects on WI-38 embryonic cells. The technology of this experiment is the result of a remarkable effort performed by scientists and NASA engineers. However, the funds for this sophisticated equipment were granted at a time of great euphoria of spaceflight. Nowadays, it is unlikely that any national or iternational agency will commit itself to financing an expensive equipment for only one experiment. In addition, WI-38 cells were attached to a substrate; therefore, possible effects of O x g on a dynamic cell-cell interaction as well as specific differentiation processes which are not taking place in WI-38 cells could not be investigated. Finally, cells anchored to a substrate forcefully assume a flat configuration regardless of the g-environment. As pointed out above (Sect. 2.2), a correlation has been found between growth and cell shape. Cell shape itself can be influenced by the g-environment. In conclusion, there are reasons indicating that cells growing in suspension and the possibility to induce them to differentiate are better objects for investigations in space.

3.3 Separation of Cells by Electrophoresis Bioprocessing of living cells is conditioned by the availability of homogeneous cell populations. The separation of specialized animal cells is one of the challenging problems of biotechnology. Since the electric charge localized on the membrane is a characteristic property of each cell, it is, in theory, possible to purify cell populations

14

A. Cogoli, A. Tschopp

by means of electrophoretic techniques. However, cells in suspension rapidly sediment to the bottom of the culture chamber and, on the other hand, Joule effects generate turbulences in the separation chambers by thermal convection. The disadvantages of thermal convection and sedimentation are ruled out in a weightless environment. Therefore, preliminary attempts to achieve separation of biomaterials by electrophoresis at 0 x g were performed in the Apollo 14 and 16 missions 2 5 ) . On Apollo 14, electrophoretic separation was successfully achieved for the first time in space with a mixture of D N A , hemoglobin and dyes. In spite of a number of problems encountered during operations, mainly due to bacterial contamination and too long storage, separation of dyes was achieved. The second attempt on Apollo 16 was less successful since the electrophoretic column leaked. The sample consisted of polystyrene latex beads of different sizes as a model for living cells. The scientific and technological importance of this method stimulated further development of the electrophoresis equipment: Two projects, one developed in the USA, the other in Germany, were flown on the Apollo Soyuz Test Project (ASTP) in 1975. In the two following sections, the experiments are described in detail. 3.3.1 Static Discontinuous Electrophoresis This experiment was developed in the USA and carried the number MA-011. R. E. Allen was the principal investigator 2 6 ' 2 7 ) . It was decided to reuse those parts of the Apollo 14 and 16 experiments which worked correctly and to change the design of those parts which failed. Bacterial degradation was avoided by using sterile procedures. Two techniques were tested: zone and isotachophoresis. The biological samples were selected according to the following criteria: a) Glutaraldehyde-fixed rabbit, human and horse red blood cells (RBC) provided a material resistant to mechanical stress and to hemolytic agents. The bands of separated RBC are easily distinguished on photographs. b) Human lymphocytes offer an ideal system consisting of several subpopulations. The separation of immunoglobulin-producing B-lymphocytes from lymphokinessecreting T-lymphocytes is a very important task for immunologists and cell biologists. In addition, a variety of T-lymphocyte subpopulations has been recently identified. The electrophoretic separation of lymphocytes, together with immunological techniques are the most promising approaches since other separation procedures such as sedimentation velocity are not applicable due to the almost identical morphology, density and size of lymphocyte subpopulations. c) A third sample contained human kidney cells. The separation of urokinase(UK)producing cells would be a very important biotechnological achievement. Urokinase is an enzyme converting plasminogen to plasmin, an obligatory step in the lysis of blood clots. Therefore, U K is an important thrombin-preventing agent. The demand for U K exceeds production capacity. UK is produced in vitro by human fetal kidney cells. However, only 5 % of the cells isolated from the cortex of the kidney by trypsinization produce UK. Consequently, the production costs of U K can be reduced by isolating and culturing UK-producing cell subpopulations. As explained above, electrophoresis could provide optimal conditions for a successful separation. The electrophoretic unit flown on the ASTP mission weighed 13.61 kg, measured 30.3 x 20.3 x 40.6 cm. By means of a self-contained power supply, requiring an external

Biotechnology in Space Laboratories

15

208-A AC, three phases and 400 Hz connection, the voltage could be varied between 0 and 99 V. Two current levels could be selected at 4.0 and 1.31 mA. A thermoelectric unit controlled the temperature between —40 and + 5 °C with the purpose of cooling or freezing the electrophoretic columns. The columns were illuminated by fluorescent light during the experiment and the evolved gas was removed by a phase separator. The separation of RBC bands was recorded by a 70 mm electric camera. The eight electrophoretic columns consisted of pyrex glass tubes of 0.635 cm inside diameter and 15.25 cm length. They were split lengthwise and rejoined by a silicone seam to compensate the expansion of the buffer solution during freezing after electrophoresis. The biological samples were contained as frozen slides in 0.13 mm thick tetrafluoroethylene cover. The slides were stowed before launch in a cryogenic canister and cooled with liquid nitrogen. They were removed immediately before the experiment was started and inserted in the separation column. The phosphate buffer (pH 7.3) was supplied by a pump. After the separation was completed the columns were frozen by operating the thermal control unit and subsequently stowed in the freezer until recovery. The inflight operations required minimum crew operations. In the ground laboratory, the frozen buffer containing the processed samples was separated from the column and sliced in 0.5—5 mm long sections. The lay-out of the samples was the following: Columns 1, 2, 3, 5, 6, 7 were used for zone electrophoresis, 1 and 5 contained fixed rabbit, human and horse RBC, 2 and 6 human lympocytes, and 3 and 7 human kidney cells. Columns 4 and 8 were used for isotachophoresis and contained fresh rabbit and human RBC, fixed rabbit and human RBC, respectively. The equipment met with expectations since electrophoresis in a space environment has been demonstrated to be feasible. However, the fluid line of columns 2, 5, 6 were clogged during pre-flight assembling causing accumulation of chemical and gaseous products in the electrode region; this resulting in voltage and p H fluctuations. In addition, column 1 was damaged during the slicing procedure. The photographs taken in the flight during the experiments showed that separation of RBC was achieved satisfactorily in column 1. Unfortunately, both columns with the lymphocyte probes failed. However, subsequent analyses revealed that the experimental conditions did not affect their viability, morphology and ability to secrete products. This is a very encouraging observation in view of future applications. The separation of kidney cells in columns 3 and 7 gave the most satisfactory results. Thus, UK-producing cells accumulated in one band and were recovered viable. This success will undoubtedly stimulate further technological developments in space laboratories. For various reasons, the isotachophoretic experiments with columns 4 and 8 did not run for a sufficient time to permit visualization of intercompartmental boundaries. However, it could be shown that this approach offers the advantages of sharp boundaries and of enrichment of the migrating zones. Proper spacers have to be developed for the fractionation of living cells. In conclusion, the MA-011 experiments demonstrated the applicability and the great advantage of electrophoretic techniques in space for basic and applied research. An improved equipment to be flown aboard the Space Shuttle is presently under development in the biotechnological units of NASA at Johnson Space Center in Houston (see Sect. 6).

16

A. Cogoli, A. Tschopp

©

© k»V*»VÎt

e

-

» • •• .•«_••• •« «1.• *

..... •.•.•.%•;.•.•.: * * * * * * * * *

, «V

.....

B Fig. 4. Principle of static, discontinuous (A) and of free-flow continuous electrophoresis (B), 1 : Flow direction of buffer with samples, 2 : Collection of fractions

3.3.2 Free-flow Continuous Electrophoresis The principle of the technique is shown in Fig. 4. The great advantage of free-flow electrophoresis is that theoretically an unlimited amount of material can be continuously processed. In addition, several samples can be easily inserted in sequence for separation. The disadvantage is an inevitable band broadening at 1 x g and at 0 x g as well. The heart of the apparatus developed by H a n n i g 2 8 2 9 ) i s the separation chamber consisting of two parallel plates isolated by a gap of 4 mm (Fig. 5). The electric field

Biotechnology in Space Laboratories

17

Fig. 5. Free-ilow continuous electrophoresis equipment flown on the Apollo-Soyuz Test Project (from Ref. 29} )

was generated by 180 mm long electrodes on the sides. The electrodes were shielded from the buffer in the chamber by ion exchange membranes. The electrode compartments were purged from gases by streaming buffer driven by a peristaltic pump. The gases generated in the buffer solution were adsorbed onto palladium. During operation, the plates were cooled at 5 °C. The buffer was driven by another peristaltic pump from the container through the separation chamber upstream the sample inlet. In this way, a constant laminar flow was produced. The buffer and the sample entered a waste container after passing the separation chamber. The samples were stored in a refrigerator at 0—5 °C. The separation was evaluated by an optoelectronic device. Totally, 4 samples were run: a) Rat bone-marrow cells, b) human and rabbit RBC, c) rat spleen cells, d) rat lymph-node cells supplemented by human RBC as a marker. A satisfactory evaluation of the separation was hindered by a malfunction of the light for optical detection. Only the zones with maximum absorption could be detected by the optoelectronic device. Nevertheless, the difficulties could be circumvented and satisfactory separation curves could be obtained. The results show that with the bone marrow cells an excellent separation was achieved. The resolution was much sharper than the corresponding separation realized in a control experiment on the ground. Conversely, insufficient information was available from the second sample, a mixture of human and rabbit RBC. Spleen cells gave the best result; several cell populations were separated and concentrated in sharp bands. In the case of sample number 4, a good resolution of lymph-mode cells was achieved at 0 x g as well as in the 1 x g ground control. Again, experiment MA-014 demonstrated the advantages of separating animal cells by electrophoretic techniques at Oxg. A further development of the apparatus described here is being prepared by Hannig to be used aboard the Spacelab (see Sect. 5.2.3).

A. Cogoli, A. Tschopp

18 SOIL SAMPLE

1 DUPLICATE TO CONTROL

ANALYSIS

LABELED NUTRIENT _ N MEDIUM r ~ ~ \

LABELED RELEASE

UNLABELED NUTRIENT, MEDIUMl

UNLABELED

§JIP

He Kr CO,

G A S EXCHANGE

Fig. 6. Diagram of the Viking experiment searching for primitive life forms on Mars. Samples of Martian soil were tested for (from Ref. 3 0 ) ): A : Synthesis of organic matter from labeled C 0 2 , B: release of labeled gases from labeled nutrients, C: change of the gas composition by metabolic action.

3.4 The Viking Mission to Mars After travelling nearly one year through the solar system, Viking 1 set down on Mars surface on July 20, 1976, followed by Viking 2 on September 3, 1976. In addition to instruments for soil and atmosphere analysis, Viking carried a sophisticated automatic miniaturized laboratory for search of primitive life forms on the Martian soil 3 0 ~ 3 3 ) . Three independent biological experiments were performed on soil samples: a) Pyrolytic release experiment (Fig. 6A): The sample was incubated for five days in a chamber under martian conditions in the presence of 14 C-labeled carbon dioxide and carbon monoxide. Artificial sunlight for photosynthesis was provided by a lamp. After incubation, all organic materials were pyrolyzed in two steps at 625 and 700 °C. Radioactivity was measured by a detector. In a control experiment martian soil was first sterilized at 170 °C prior to incubation and pyrolysis.

Biotechnology in Space Laboratories

19

b) Label-release experiment (Fig. 6B). Soil samples were incubated in the presence of nutrient medium containing 14 C-labeled formate, lactate, glycine, alanine, and glycolic acids. Living microorganisms, when present, would release labeled carbon monoxide and carbon dioxide or methane. Again, a control sample was sterilized prior to incubation. c) Gas-exchange experiment (Fig. 6C). Martian soil was incubated with unlabeled medium in a controlled atmosphere of He, Kr and C 0 2 for 7 days. Gases like hydrogen, nitrogen, oxygen, methane, and carbon dioxide eventually released by microorganisms were measured with a conductivity detector. Initially, the experiments delivered results which were consistent with the presence of life forms on the martian surface. However, the chemical analysis of soil samples revealed no organic compounds larger than methane and propane ^ A possible explanation of the puzzling results is that the reactions observed may be catalyzed by peroxides (y-Fe 2 0 3 ) present in the surface of the planet. In conclusion, this exciting project did not give a definite answer to the question whether there is life on Mars or not. Further experiments which must take into account the peculiar composition of the Martian soil are required to solve the problems. However, simulation experiments conducted on earth make a biological interpretation of the Viking results most unlikely 34 • 35) .

4 Biological Payloads on Soviet Missions The material presented in this section has been selected f r o m : a) translations into English from Russian of original articles, memoranda and bulletins, mainly from the Journal of Space Biology and Aerospace Medicine which is a translation (edited by NASA) of Kosmicheskaya Biologiya I Aviakosmicheskaya Medsina, Moscow; b) original reports published by the Soviets mainly in the COSPAR (Committee on Space Research) series Life Sciences and Space Research. A comprehensive collection of literature references concerning soviet experiments appeared recently 3 6 ) . A great number of biological experiments was performed on manned and unmanned Soviet satellites. Unfortunately, very few technical details of the hardware have been published. The Soviet space program in life seciences was concerned with the study of adaptation of man in space, basic research of the effects of 0 x g and cosmic radiation on microorganisms, plants, seeds and seedlings, embryos, mammals (mainly rats) and, interestingly, with the development of autonomous greenhouses in space. We can divide the activities of the Soviets in three phases: a) The pioneering phase of manned spaceflight, characterized by the missions of the spaceships Vostock I—VI and Voskhod I and II (1961—65). Already cosmonaut Gagarin carried on Vostock I containers with E. coli and other microorganisms 37) . b) The second phase is marked by the launch of automated biosatellites of the series Cosmos (Table 2) containing highly sophisticated equipment like life-supporting systems for rats and centrifuges generating 1 x g conditions. c) The era of a systematic exploitation of space resources began for the Soviets in 1971 with the launch of orbital stations periodically visited by crews travelling on Soyuz vessels and refurbished by unmanned space freighters of the type Progress. The launch of the Salyut-1 space station was followed by Salyut-3 (1974), Salyut-4

A. Cogoli, A. Tschopp

20

e

e

o U

C3 -o Lh cS X

o o. a 3

o a Q. 3

C O u

ai

•U E u O Vi 3 t/> O Vi 60 C O O 'S D, -O a. a 3Vi Ih VI u J 2

¿

S o.

60 u

C 60 •5 a o 'C ri ir G 3 u ¿Î 60 X J —

XI o

o 3 o S

o t: o

60 3

o

s

so o TJ O S H

•a c CS vi

3 •o

a. c» . t. ca c o O SO' °

u

2 a o

=3 C 13 « o =5!

o Si o7 O U vi

J3 60 s 32 O 1 00M o •» Ig s 0

2 o 3 e .ti o > 2f 0 c U 00 I J* < o Ü O X • e I D ° ' 8 2 => _o i C --1 S " r a s 's "o ! « ^ cd O - J3 i • 2 II c o J= S i M . >> Ë D. !1 00 S " O u 3 I -o c «e s V V O I - c S > ° r _ I 00 cd "> £ es ¡3 e L, y O - Ü0 J3 ! ° g « 'G «> ¡ 4) ~ tn c a » ' -o c / D a -o u X 3 v> w c c jj« "Ü U u a § g V) C Où rC c c O í0> «ì 2 ¡2 -o S-o CA g .2 « 7 ^ 2 x> ^ ! -O «^ c 2 « 3 u g 00 o ca u "O CA-S X l í Ï "Öj ). Here we briefly describe the main facilities offered. The module is made of cylindrical shells and the pallet of segments (Fig. 8). The length of the module and that of the pallet can be varied according to the requirements of the mission and extended either to an "all-module" or to an "all-pallet" set-up. The hardware used for the experiments is fixed to racks in the module and on the ground of the pallet. The total weight of the payload ranges from 5.500 kg for a module/pallet to 9.100 kg for an all-pallet set-up. The power available is either 28 V DC or 115/200 V 400 Hz, AC. The environmental control provides the module with an air temperature of 18—26 °C and a relative humidity of 30—70 %. The orbit altitude ranges between 200 and 900 km. A typical flight profile of a Spacelab mission is given in Fig. 9. A selection of the experiments is made periodically by ESA and NASA on the basis of proposals submitted by investigators in answer to announcements of flight opportunities. In addition, the agencies are ready to evaluate the scientific relevance and the technical feasibility of proposals submitted at any time.

5.1 The First Spacelab Mission The objective of the first Spacelab Mission is to show the scientific community that the facilities offered are adequate to perform basic and applied research taking profit of the peculiar space environment. It is a multidisciplinary mission carrying experiments from Europe, USA, India and Japan. Great efforts were made by the

27

Biotechnology in Space Laboratories

EXPERIMENTER SPACEFLIGHT TRAINING

MCC/POCC

OPF O & C TDRSS 6

Fig. 9. Shuttle/Spacelab mission cycle (from Refs. °.

6163

>)

Mission Control Center Payload Operations Control Center Orbiter Processing Facility Operations & Checkout Tracking and Data Relay Satellite System

agencies and investigators to accommodate all instruments overcoming the constraints of weight, room, energy, and crew time. For the same reasons, experimental equipment and operations should be designed to minimize the use of the facilities available. Therefore, severe limits are set to the experimental resources and relatively simple experiments have a higher probability of success. Better conditions will be offered by the follow-on Spacelab program with the advent of discipline-dedicated flights. The life science projects of Spacelab-1 include physiological investigations on humans (crew members) and biological experiments on plants, microorganisms and cells in vitro70). The vestibular system and the space motion sickness in man will be investigated in experiments 1ES201 from Germany, 1NS102 and 1NS104 from USA. Blood specimens will be withdrawn inflight and analyzed later on the ground for red cell mass reduction (1NS103, USA), hormonal levels (1ES026 and 1ES032, Germany) and immunological parameters (INS 105, USA). Periodic motions of the human body, resulting from heart activity and blood ejection into the aorta, will be recorded by three-dimension ballistocardiography (1ES028, Italy). Several physiolo-

28

A . Cogoli, A .

s •o" ' o a C l-l 60 cd X 60 o X cd — •o « e •S

co c

a 'S

U a tA O O öl « c O.— O CA 4> 5 6 0 "o S O e .a -e E ts

t_i -C b

i

s

1 . 1 •O O

l - s 7J ü CU Q

•a 3

S S. S c«

C

.2 tt -t-> U « 3

e 1 (D > . c -S >" I 2 ? a t-i üo 60 ^

' os e O S 3 E

a o

"O Vh •o S •S i-

6o CS

M o

cd O. c o

O S •P

g >

eg '5 nt -a « t« E »

«u

"O ed kH O

JS "

c T: O ci

"3

•8 o o ti o. g X o, u .. S

c

o J3 H Q. E ûû >J s

" E

cd c C/2 U

a. E cd

i o

8 o. cd c«_ Oh O

'S « tü

~3 •a o S

O E 5

cd e

73




en 2 s

¡ . a

s i - s • 0 ^ 3

• S I ' S S • rf LH u O x> a D C t: ^ cd o cd o. o £ . .O g s rG

o J3 G. E

c cd E ~ c 3 K -ci « J3 O 3 Cfi a _ J .S3 N o c m

E — cd

& o

o

S ^

m

o. X w

U O O «M o O l-l a. E

c

s

60 O "Ö

o • Vi (A C/}

> o

X -s m £ .S

U

V 'M hjû C c S

'S

l ' I ' S OD 60 '3 «

c o

•S O ^P "S % u VIfc en E U u C/3 O >. u M ' a O

I ° S J3 S o. CA~ OJ T3 00 . a» 3 O cC » • -

i>

cd

S .2

=3 « Û, V

S I o X) T3 C*H • o O O

s i

1 CA O 15

.2

à » e c aA o C -O Ë ë

^ 3

2

a ^ > e cd E

< t/3

< C/3

S

P

O

Ü

C/0 Z

8) or from bubbles bursting from the microbially laden surface films of rivers, lakes, and oceans 9) , or from spray generated by breaking waves 1 0 , 1 1 ) and rainwater splashes 12>13) . Dislodgement from vegetation or soil as result of wind action or thermal convection 3 ' 1 4 ' 1 5 , 1 6 , 1 7 ) is another possible source. People also disperse microbes by talking, coughing, sneezing 1 8 , 6 ) and by the shedding of skin 3 ' 6 , 1 , 9 ) . Persons capable of shedding microorganisms from the respiratory tract or other areas of the body in abnormally high numbers (a rate of Si 100,000 viable particles per minute, Kraidman 1 9 ) are termed microbial 'shedders' and present a constant source of air contamination. It has been estimated that airborne microbes of 1 |im diameter could be carried 14,000 km (9,000 miles) in a steady 16 km h " 1 (10 m.p.h.) wind before settling 92 m (100 yards). Bacteria have been isolated at altitudes as high as 256 km (16 miles) 2 0 ) to 768 km (48 miles) 21) . Evidently, a generator of pathogenic microbes could exert its effect (contamination or disease) thousands of kilometers away with no apparent link between the terminal settling area and the source. However, the infectivity and viability of airborne microorganisms, carried by the wind, is limited by the availability of substrates 22) .

-

0.1

Fig. 1. Aerosol stability of Serratia marcescens as a function of culture age (h). (Figure after Goodlow and Leonard 4) ) 10

12 14

Culture age, h

53

Basic Concepts in Microbial Aerosols

The size of the particle is dependent upon, among other variables the type of microbe, the chemical content of the fluid from which the droplet nuclei originate, as well as the relative humidity (RH) of the atmosphere in which they are suspended 23) . If a particle is less than 10 |am in size, it is defined as a droplet nucleus 2 4 ) . Particles over 10 |im in size are almost completely screened out in the upper respiratory passages of man, along with 80% of particles 5 nm in size. Smaller sized particles are inhaled and preferentially deposited in the lung 2 5 ) , or, if not inhaled, tend to remain suspended in confined and inhabited space until vented or inhaled 2 6 ) . Dimmick and Akers 11 state that as droplet size decreases, the surface exposed to air resistance becomes relatively greater when compared to weight or gravitational attraction towards the earth. This ability to remain in static air constitutes a biological hazard in areas of unvented air. Persons having open wounds, boils, or carbuncles shed microbes of the genera Streptococcus, Klebsiella and Pseudomonas by air 19) . Other airborne diseases and pathogenic microorganisms are tabulated in Table 1 27) . The American Association for Contamination Control in 1965 established air quality requirements for hospital rooms, burn areas, dressing rooms and nurseries. F o r a 60 minute continuously sampled period, the flow of air entering these areas should contain no more than one particle-bearing, colonyforming aerobic bacteria per cubic foot; the air already present within this space should not hold more than 5 particles per cubic foot. However, air conditioners, ventilators, fans, etc. can create aerosols and disperse them throughout a room thereby increasing the probability of contagion inhalation or deposition. The parameters involved in the assay as well as viability and techniques are reviewed in the following sections. Table 2 summarizes some of the work done in aerosol viability studies. Table 2 lists much of the work in aerobiology identifying and elaborating on the stresses involved in microbial aerosols. The other factors that are probably involved in bacterial aerosols are the physical forces associated with the generation and recovery of the aerosol and the presence of extraneous material such as spent culture fluids or tissue exudates in the aerosol or collecting fluids 2 8 ) . The increase or decrease of the death rate of microbes in these aerosols depends upon the quality and/or quantity of the above factors.

Table 1. Pathogenic microorganisms and diseases caused by air transfer Organisms: Adenoviruses Bordetella pertussis Cryptococcus sp. Diplococcus pneumoniae Mycobacterium tuberculosis Mycoplasma pneumoniae Neisseria meningitis Poliomyelitis virus Rubeola virus Streptococcus haemolytica Variola virus

Diseases: Blastomycosis Histoplasmosis Influenza Mumps Psittocosis

54

J. E. Zajic, I. I. Inculet and P. Martin

Table 2. Factors involved in microbial aerosol viability Parameters

Author

Ref.

Bacterial strain

Goodlow & Leonard (1961) Dimmick & Akers (1969) Pelczar & Reid (1972) Goodlow & Leonard (1961) Dimmick & Akers (1969) Cox (1971) Goodlow & Leonard (1961) Hayakawa & Poon (1965) Akers et al. (1966) Hatch & Dimmick (1966) Cox (1966 a, b) Anderson (1966) Wright et al. (1969) Dimmick & Akers (1969) Ehrlich et al. (1970) Ehrlich & Miller (1973) Cox et al. (1974) Dimmick (1960) Hayakawa & Poon (1965) Hess (1965) Akers et al. (1966) Hatch & Dimmick (1966) Won & Ross (1966) Cox (1966 a, b) Anderson (1966) Zentner(1966) Cox (1968 a) Wright et al. (1968) Dimmick & Akers (1969) Stewart & Wright (1970) Ehrlich et al. (1970b) Hatch et al. (1970) Southey & Harper (1971) Riley & Kaufman (1972) Cox & Goldberg (1972) Lighthart (1973) Turner & Salmonsen (1973) Ehresmann & Hatch (1975) Cox (1976) Hatch & Dimmick (1966) Anderson & Cox (1967) Hess (1965) Cox (1967; 1968 a) Benbough (1969) Cox et al. (1974) Cox et al. (1973) Cox et al. (1971) Kundsin (1966; 1968) Serat et al. (1969) Lighthart et al. (1971) Cox et al. (1973) Lighthart (1973)

*)

Incubation medium Temperature

Relative humidity

Cellular physiology Oxygen concentration

Light and irradiation

ì) 27) 4) 1) 29) 4) 2) 30) 31) 6,7) 28) 32) 1) 33,34) 35) 36) 5) 2) 37) 30) 31) 38) 6,7) 28) 39) 40) 41) 1) 42) 34) 43) 17) 44) 45) 46) 47) 48) 49) 31) 50) 37) 51,40) 52) 36) 54) 53) 23,26) 57) 58) 59) 46)

55

Basic Concepts in Microbial Aerosols Table 2 (continued) Parameters

Author

Ref.

Ion transport disruption

Anderson et al. (1968) Hambleton (1971) Benbough et al. (1972) Anderson (1966) Cox (1966a, b) Hatch & Dimmick (1966) Won & Ross (1966) Kethley et al. (1957)

60)

Sampling medium

61)

62) 28) 6,7) 31) 38) 63)

2 Assay Techniques Many nebulizers are obtainable that generate aerosols from solutions depending on the particle size, uniformity of droplets, and dispersion conditions required. Information on the types and characteristics of atomizers has been reported by Mercer M ) , Dimmick and Akers 11 and Larson 65) . Dimmick and Akers suggest a direct way to measure the output of cells from a nebulizer by straining the cells, dispersing them with distilled water, collecting them in a sampler (for further information see Table 3), and measuring the concentration by photometric methods. The volume (ml) of the dispersed droplet times the number of cells per ml yields the expected probability of a cell being in a droplet. The number distribution of aerosolized particle sizes is approximately log normal 6 6 ) except for the smallest particle size. If higher numbers of cells are being contained in the smaller particles 4) , the concentration of cells in the dispersing fluid is increased and the solid content as well as the particle density increases in the atomizer fluid. This effect is negligible for spray times less than 5 min 1). The decay of a microbial aerosol can be estimated by quantifying the number of airborne organisms per unit of air volume at various cloud ages. The slope of the resulting curve (density vs cloud age) is the rate of decay of the aerosol cloud (% per min). Included in this decay rate are physical decay caused by fallout or deposition on the chamber walls and biological decay caused by biological inactivation of the microbes 33) .

Table 3. References for the collection and sampling equipment used in aerosol studies Ref. Luckiesh, Taylor and Holladay (1946) Cown, Kethley and Fincher (1957) Houwink and Rolvink (1957) Andersen (1958) Decker et al. (1958) Wolochow (1958) Batchelor (1960)

67)

68) 69) 70) 71) 72) 73)

Ref. Andersen and Andersen (1962) Lidwell and Noble (1965) Malligo and Idoine (1964) May (1964) Mercer, Goddard and Flores (1965) Dimmick and Akers (1969) Knuth (1969)

74) 75) 76) 77) 64)

1) 78)

56

J. E. Zajic, I. I. Inculet and P. Martin

Recovery of a tracer disseminated as a mixture with the test microbes allows an estimation of the physical decay of the aerosol. Therefore, the difference between the total decay rate and the physical decay rate provides an estimate of the biological decay of the organisms. Hatch and Dimmick 31) state that, while physical decay always follows first order kinetics, the biological decay rate, depending on the conditions prevailing, only approximates first order kinetics. Cox et a l . 7 9 ) provide equations to estimate the components (electrostatic precipitation, inertial impaction, gravitational settling, and Brownian diffusion) of the aerosol physical decay. A phenomenon of the decay rate that must be taken into account during aerosolization is the extremely high decay rate in the first second of the cloud age which is followed by a slower rate of decay during the next nine seconds. This means that the decay rate is not necessarily constant with respect to time. Assuming first order kinetics, the change in the number of aerosol particles, dN, is equal to the change in time of the suspended particles, dt, times a constant rate k, and the microbe concentration N, dN i.e. — = —kN dt

(1)

Rearranging and integrating Eq. (1) from t = 0 to some time t yields n,

, d N

, j k dt

N dN

= kt

(2)

(3)

N0 Integrating and solving for k gives k = — In i — t Vn,

(4)

where k = decay rate constant, s " 1 t = time of sampling, s N c = original number (or concentration) of aerosol particles (microbes), particles-cm - 1 N, = number (or concentration) of aerosol particles (microbes) at sampling time t, particles-cm - 1 . Hayakawa and Poon 2> used Eq. (4) to derive the decay rate constant for their data. This is not the only approach to evaluating microbial activity in aerosols. For biological decay, Hatch and Dimmick 3 1 ) followed the change in relative humidity (RH) from the initial cloud conditions to the final cloud conditions and correlated the biological loss of the aerosol to a dynamic-humidity-death (DHD) ratio (Fig. 2).

Basic Concepts in Microbial Aerosols

1.0

57

Initial humidity condition Final humidity condition

— 0.1 -

-

. Apparent dilution ratio, 1/0.59 (ADR)

Biological l o s s r ^ . initial 70% 1 Biological "—•— £dilution_raj[o_0.59 JBDR) Biological loss, - \ f i n a l 70% \ J With no change in humidity DHD ratio = 1.00

o,0.01 - 1.0 Apparent dilution ratio 0.51 Biological loss," initial 70% 0.1 _

Humidity shifted at this time at the confluence point With shift in humidity DHD ratio = 0.77

0.01

[• Biological dilution ratio 0.55 \

\

Biological loss, final 90

\

Physical Biological

5.7 Aerosol time (mir)

Fig. 2. The change in survival rate with a change in the relative humidity of suspended bacterial aerosol particles. The physical loss is due to natural forces (e.g. coagulation, deposition, etc.); the biological loss refers to the loss of viability. (Figure after Hatch and Dimmick 31))

11.3

Other quantitative methods relate the colony forming units (CFU) of the experimental microorganisms to the CFU of a microbial tracer (spores) to yield the percent viability (or percent survival; Cox and Goldberg 4 5 '): , viability =

CFU (test organism) CFU (tracer organism)

100*

(5)

Lighthart 46) employed a graph of —log (% survival) versus time interval and states that the death rate is the negative slope of the line, fitted by the method of least squares through the experimental points. The equation obtained from this is survival =

total decay physical decay (NM/NUT) (NBT/NB.)

x 100

x 100

(6)

* The ratio of CFU's was normalized to the initial CFU ratio in the aerosol medium to obtain a comparative reference point to 100% viability.

J. E. Zajic, I. I. Inculet and P. Martin

58

where N u = number of CFU of the test organisms in the initial sample N u t = number of CFU of the test organism at time t N b ! = number of tracer spore CFU in the initial sample N b t = number of tracer spore CFU at time t. The previous equations were used to describe the "life-times" of bacteria in the aerosol state. Other algebraic expressions that deal with aerosol viability are mainly concerned with the effect of one or more factors involved in the death rate. An equation for the oxygen effect on viability was developed by Cox et al. 36>, one for a denaturation effect by Cox 49) , and one for exposure of bacteria on microthreads to an open air factor effect (the action of pollutants probably formed from ozone and olefin interaction) by Cox et al. 5 9 ) . All of these formulations give a good approximation to the data. They are only approximations and may not be comparable to a specific measurement of a specific microbial aerosol. Table 4 includes the tracer material used for assay and also some references to authors using each method. All of these techniques have certain inherent disadvantages. Chemical tracers have the following difficulties associated with their use. Tracers of relatively small molecular weight (e.g. sodium fluorescein, Rhodamine B or a

Table 4. Assay tracers and experimental use Tracer or Isotope

Author

Ref.

Dyes

Dunklin & Puck (1948) Henderson (1952) Wolfe (1961) Dorsey et al. (1970) Harper & Morton (1952) Harper et al. (1958) Harper (1963) Harstad (1965) Green & Green (1968) Benbough (1969) Miller et al. (1961) Green & Green (1968) Anderson (1966) Benbough et al. (1972) Strange et al. (1971) Strange & Martin (1972) Strange et al. (1972) Anderson (1966) Cox (1966a, b; 1968a, b; 1971; 1976) Benbough (1969) Ehrlich et al. (1970) Hambleton (1970) Strange et al. (1972) Cox & Goldberg (1972) Ehrlich & Miller (1973) Lighthart (1973) Cox et al. (1974)

81)

32p

35

S

14C 125j

Bacillus subtilis var. niger spores

82) 83) 55) 84) 85) 86) 87) 88) 52) 89) 88)

28) 62) 90) 91) 92)

28) 6,7,40,93,29.49) 52) 34) 94) 92) 45) 35) 46) 36)

Basic Concepts in Microbial Aerosols

59

radioisotope) are evenly distributed within the aerosol medium and particles. However, unless the tracer is specific for the microbe, the recorded count will vary because some of the aerosol particles do not contain microbes. In other words, the difference in the physical properties of the particles affects the collection ratio of tracer to microbial particles. Radioactive tracers have the same difficulties but, in addition, safety must be considered. Viability problems are encountered in the use of spores as tracers. There is evidence that a proportion of the spore tracer may die in the aerosol 8 5 , 8 0 ' 2 8 , 7 , 3 4 ' ) . Strange et al. 92) have been shown that about 10 % of the initially viable spores (Bacillus subtilis var. niger) lost viability or became dormant under certain conditions in an aerosol. This loss of tracer CFU ability imparts a higher survival or viability percentage and a lower death rate to the test microbes than is warranted. Overall, assays of different bacterial aerosols generally have shown that Gram negative microbes are more sensitive to aerosolization than are Gram positive organisms 9 5 , 8 0 ' 9 6 , 8 8 ) . Goodlow and Leonard 4 ' state that thermophilic mutants of several pathogens could possibly show increased resistance to aerosolization. Finally, despite the great physiological and structural differences between the true bacteria, mycoplasmas, and L-forms, these species show similar responses to the stress of being in the airborne state 2 6 , 4 1 ' 3 2 ' 4 2 ) .

3 Concept of Cell Ion Units The permeability of some Gram negative bacteria is affected by aerosolization. Lack of control of ion transport 6 0 ) and increased permeability 9 8 , 9 9 ) have been associated with microorganisms recovered from aerosols. In addition, there is a change in the outer layers of the cell envelope 94 '. Consequently, organisms which have been recovered from aerosols should not be regarded as unchanged rehydrated forms of the original bacteria. Permeability changes resulting from aerosolization may not be lethal in themselves but could be directly or indirectly related to the subsequent survival of airborne bacteria 6 0 ' 9 4 , 6 1 ) . Damadian 1001 hypothesized that E. coli can be viewed as simple, multiphase ionexchange systems and that the movement of ions such as potassium ( K + ) and sodium (Na + ) into the cells during growth and respiration are only counterions for newly formed polymers or acid products of catabolism. However, Anderson and Dark 101 ' think that damage to a K + retention system is a consequence of the aerosolization of E. coli. In fact, it is the general opinion 2 8 , 1 0 1 , 5 2 , 9 4 ) that K + retention is lost by most aerosolized bacteria. Aerosolization of E. coli decreases their ability to synthesize certain macromolecules and concentrate ions 6 0 ' 5 2 , 6 2 ) . Anderson et al. 601 state that the loss of control of K + retention results in a general disorganization of the bacterial ion and substrate transport mechanism. However, the increased K + permeability can be counteracted by the addition of K + in the growth medium. Anderson 28) established that populations of E. coli B showed a severe decrease in the ability to synthesize protein immediately after aerosol sampling. Since there is evidence that ribosomal function depends on K + in a mutant of E. coli and in Aerobacter aerogenes103), the decreased rate of protein production could be due to

60

J. E. Zajic, I. I. Inculet and P. Martin

the loss of K + control. Inasmuch as K + is a cofactor for many enzymes in carbohydrate and aromatic compound metabolism 101) , these reactions could also be affected by a disturbance in the K + retention system. Benbough et al. 62) found that aerosolization caused a detachment of bacterial membrane components. The transport mechanism was restored by incubation of the microbes with the lost components. Increased K + permeability and the loss of incubation protein synthesis are also thought to be reversible 28 • 1 0 4 , 1 0 1 \ On the other hand, experimental work with 32 P has illustrated that much of the loss of the phosphate ion from the cell has been due to procedural methods previous to aerosolization 101). It appears that the process or procedure for atomization or a stress involved in being suspended in air can cause a partially reversible and partially lethal loss of at least one ionic component in aerosolized bacteria.

4 The Effect of Air Composition on Microbial Aerosols 4.1 Oxygen The effect of oxygen on airborne organisms are complex and could either be toxic or protective. A protective action of oxygen was hypothesized to occur through a suppression of the activity of a toxic component in the spent culture suspending medium 29) . Conversely, if Serratia marcescens were held in an oxygen free atmosphere, there was little or no viability loss 37) i.e. the toxic effect of oxygen was absent. There appear to be similarities in the toxicity of oxygen towards aerosolized bacteria and its effect on enzyme activity 105>, irradiated organisms 106 ' 107 • 108) , and freeze-dried organisms 109 - 110) . Cox et al. 3 6 ) state that aerosolization and freeze-drying can both result in the same oxygen induced death mechanism. They derived a kinetics equation for the viability of freeze-dried bacteria to fairly accurately describe the survival of aerosolized bacteria in varying oxygen concentrations. Cox and Baldwin m ) suggest that the oxygen toxicity towards aerosolized bacteria may be a free-radical induced phenomenon. Freeze-dried Escherichia coli have shown electron spin resonance signals u 2 , 1 1 3 , 1 0 9 ) which are indicative of free-radical formation. These signals were greatly enhanced on exposure of the freeze-dried microbes to oxygen. Since oxygen induced free-radicals decay rapidly in the presence of moderate amounts of water vapour 114), this toxicity can occur at low relative humidity (e.g. low survival of Pasteurella tularensis LVS at low RH values, Cox 2 9 '. If oxygen toxicity occurs at high and intermediate RH values (see Table 2) 6 , 4 0 , 5 2 ) another process must be involved in the death mechanism. Oxidation by peroxy radicals (from exposure to oxygen) at low RH is a consequence of bacterial dehydration. In other words, the lethal action of oxygen appears to be associated with the loss of water from bacteria. Oxygen was shown to be progressively more toxic as the humidity is reduced below 65% 5 2 ) . Bateman et al. U 5 ) pointed out that perhaps it is the strongly bound water that is lost when bacteria are exposed to RH values of 70 % and less. The structural conformation of bound water could

Basic Concepts in Microbial Aerosols

61

alter oxygen accessibility to the bacteria and its reactivity with bacterial macromolecules 80) . Oxygen has been found, at atmospheric pressure, to produce enzymatic inhibitions 116) . Specifically, hyperbaric oxygen oxidizes the dithiol moiety of a-lipoic acid, a cofactor of pyruvate oxidase which, in effect, decreases enzyme activity in brain cells 117) . Barron U 8 ) and Hangaard et al. 105) have also observed the toxic action of oxygen on pyruvate oxidation in gonoccoci and heart homogenate, respectively. Although it has not been demonstrated, this action of oxygen on enzyme systems could occur in aerosolized bacteria. The loss of the outer membrane components of aerosolized Gram negative bacteria by the process of aerosolization, as described previously, could expose microbial proteins to the toxic effect of oxygen. The sensitivity of the microorganisms in aerosols would therefore increase. Cox 4 0 , 2 9 ) suggests that the toxic action of oxygen is on the inhibition of cell division, or cell synthesis, or both. The site of the oxygen reaction could be in the periplasmic gap 3 6 ) . Goodlow and Leonard 4) found that minimal bacterial survival in air is obtained with rapidly metabolizing 5. marcescens. Gershmann et al. 1 1 9 ) also noted that protection against oxygen poisoning is the greatest during decreased metabolism of irradiated microbes. Oxygen toxicity can be altered by changing the growth conditions or by including certain agents (e.g. Mn + 2 , Co + 2 , glycerol and thiourea) in the spray fluid. These agents also increase the viability of freeze-dried bacteria 52) . Physical forces, water content, metabolism, medium constituents and oxygen concentration are the variables that should be included in any study of the effect of oxygen on microbial aerosol survival.

4.2 Other Gases Bacterial aerosol survival in air can be similar to (at high RH) or much less than (at low RH) survival in pure nitrogen 7). Freeze-dried organisms also show an increased survival in atmospheres of nitrogen, hydrogen, and carbon dioxide when compared to air or oxygen 120,121) . However, Webb 80) found that the viability of aerosolized bacteria was not different in air than in other gases. Carbon monoxide can have a protective or lethal effect on aerosolized bacteria in concentrations present in the urban environment. The death rate of Serratia marcescens and Sarcina lutea was enhanced by carbon monoxide (85 jj.1 per 1 air) at low RH. Cell protection was shown at high RH values. The microbial species, aerosol age, and RH are some of the factors that have been studied by Lighthart 4 6 ' in conjunction with carbon monoxide as a contaminant. In some organisms, cellular metabolism is affected by a carbon monoxidecytochrome reaction resulting in photosensitive carbonyl compounds 1 2 2 ' 1 2 3 ) . It has also been shown that bacterial chromosome replication depends upon a carbon monoxide sensitive reaction which in itself affects bacterial survival 124) . Lighthart 4 6 ) explained his results from the viewpoint of the inhibition of an energy-requiring death mechanism at high RH and an inhibition of energy necessary for cell maintenance at low RH values. Lighthart et al. 5 8 ) used sulphur dioxide as a variable and concluded that the death rate of S. marcescens increases directly with gas concentration and inversely

62

J. E. Zajic, I. I. Inculet and P. Martin

with RH. The open air factor (ozone-olefin interaction), in the parts per hundred million range, is very detrimental to microbial aerosol survival 59) . Cox et al. 5 9 ) advocate the use of bacterial sensitivity as an indicator of the concentration of contaminants. Other pure gaseous atmospheres of nitrogen, argon, and helium have been used in aerosol studies to exclude the detrimental action of oxygen on the airborne microbes. Cox 4 0 ) found, at high RH, that these gases were biologically active and possibly formed clathrates with cellular water. These interactions will be discussed briefly in the section concerning RH.

5 Cell Surface The cell surface is affected by almost all aspects of the aerosolization process (e.g. surface charge, binding and adhesion, amount of bound water, reaction with inert air pariculates, etc.). Most bacterial and plant cells are surrounded by a wall, consisting of a thick porous, polymeric meshwork that may occupy as much as 50%

D-glu

D-ala Fig. 3. Schematic representation of a general bacterial cell wall peptidoglycan complex. The polysaccharide chains are covalently cross-linked to each other via small polypeptide bridges. In this diagram the bridge is composed of a pentaglycine chain joining two tetrapeptide side chains. (Figure after Bohinski 131 >)

Basic Concepts in Microbial Aerosols

63

of the total volume 125) . The surface structures of a variety of bacterial cells are extensible, flexible, and elastic as shown by the study of Knaysi 1 2 6 ) . Electron microscopic studies 127) have shown that Gram negative bacterial cells consist of an inner rigid peptidoglycan layer and a membrane-like outer 'soft' layer. The peptidoglycan moiety comprising a part of the cell wall structure can be visulaized as a 3-dimensional rope ladder with relatively rigid polysaccharide rings and relatively flexible polypeptide ropes (Fig. 3, 4 and 5). Ou and Marquis 125> describe this network as a predominantly elastic restraining structure rather than a rigid shell. Figure 5 shows the main constituents of the Escherichia coli cell envelope and the sites of action of three etizymes. The flexible lipoprotein-lipopolysaccharide component 1 2 3 ' 1 2 9 ) that overlays the layer of mucopolysaccharide (peptidoglycan) protects this layer from the action of lysozyme, possibly by impeding the penetration of the enzyme into the wall 9 4 ) . In support of this theory, Salton 128) and Noller and Hartsell 1 3 0 ) demonstrated that certain pretreatments, which appear to act by disruption of the outer wall complex of Gram negative bacteria, sensitize these bacteria to lysozyme. The effect of aerosolization on the cell envelope may be non-specific 60) . Hambleton 94> states that, following aerosolization of bacteria in air at 75 % relative humidity (RH), some Gram negative bacteria show structural changes in their envelopes, manifested in the susceptibility of the organisms to some enzymes, in particular, lysozyme.

Fig. 4. Schematic diagram of the peptide-carbohydrate repeating units (Fig. 3) repeated to yield a rigid, gridlike structure. The 'm' refers to the carbohydrate that is involved in the crosslinking (Figure after Bohinski

Lysozyme

Peptidoglycan

Fig. 5. Simplified model of the cell wall envelope of Escherichia coli. The underlined (with broken lines) compounds are enzymes and the arrows indicate the site of the enzymatic activity. (Figure after Schnaitman 132) )

64

J. E. Zajic, I. I. Inculet and P. Martin

Hambleton 61 ' found that certain di- and trivalent cations added to suspensions of bacteria recovered from aerosols mitigate this lysozyme sensitivity. Aspell and Eagon 133) and Schnaitman 132) seem to agree that the role of divalent cations is in the stabilization and organization of the lipopolysaccharide-lipoprotein layers of the Gram negative cell envelope. Therefore, since sensitivity to lysozyme does occur and divalent cations mediate a resistance to the enzyme, aerosolization must stress or structurally affect the permeability of the outer membrane to lysozyme (Fig. 5). Hambleton 61) also states that the repair of wall damage of aerosolized bacteria is similar to the restoration of permeability control occurring in organisms recovering from EDTA (ethylene diaminetetraacetic acid) treatment 134,132) . This hypothesis can be easily tested by analytic comparison of the results of the two treatments. Lieve 135) has demonstrated that the changes in permeability inducted in E. coli by EDTA can be reversed by an energy utilizing process in the microbe. Maximum wall repair of aerosolized bacteria has been shown to take place only when the microorganisms were incubated in a complete medium in the presence of multivalent cations 61 '. On the other hand, aerosolized bacteria, in the presence of 2,4-dinitrophenol, which uncouples oxidative phosphorylation 136) , are unable to restore lysozyme resistance (i.e. to repair the outer membrane; Hambleton 61) . It is reported that mycoplasmas or L-forms form viable aerosols even though they lack a cell wall 2 3 ' 2 6 ' 4 2 ) . Preferential removal of surface cell wall components is possible. In E. coli, ethanol removes the lipoprotein of the cell wall to expose the mucopeptide layer 137,138) . Enzymatic removal of the carboxyl terminal groups of surface amino acids was obtained by incubating E. coli under certain conditions and harvesting a specific decarboxylase 139) . Blockage of the amino groups on the cell surface can be obtained by treatment with hyaluronidase and fluorodinitrobenzene (Ingram and Salton l40) ). Osmotic variations of aerosolized bacteria will be discussed in the section concerning relative humidity and water effects in aerosolization.

6 Relative Humidity 6.1 General Water is an ubiquitous compound that is essential to all life. The liquid phases surrounding and dominating the inside of a bacterial cell govern the rate of transport of nutrients, ions and gases, into and out of the cell. These phases are sensitive to variations in the relative humidity (RH) of the environment. Table 5 shows that various aerosolized bacteria have ranges of maximal sensitivity to RH. A dual aerosol transport apparatus (DATA) was developed by Hatch and Dimmick 3 1 ) to adiabatically change the humidity to which airborne cells were initially equilibrated to a higher or lower humidity level and to study the effects of humidity shifts. An airborne microbe undergoing a shift in the environmental RH will have an increased death rate when compared to the death rate values before the shift. The modification of cellular biological properties depends on the direction of

Basic Concepts in Microbial Aerosols

65

Table 5. References for aerosol experimental results supporting different RH ranges of greatest inactivation Low RH ( a 60%)

Wells & Zapposodi (1948) 1411 Dimmick (I960)" 51 Goodlow & Leonhard (1961)4> Hayakawa & Poon (1965) 2) Lighthart (1973) 461 Cox, Gagen & Baxter (1974)36)

Dunklin & Puck (1948) 8l) Schechmeister & Goldberg (1950) 142) Ferry & Maple (1954) 143> Beebe (1959) 1441 Webb (1959) 80) Anderson, Dark & Peto (1968) 60) Wright, Bailey & Hatch (1968a) 41) Hatch & Wolochow (1969) 1451 Wright, Bailey & Goldberg (1969) 321 Stewart & Wright (1970) 42) Cox (1976)" 491

Williamson & Gotaas (1942) 146 ' Loosli et al. (1943) 1471 DeOme (1944) 148) Dimmick (I960)3 s> Goodlow & Leonard (1961) 4) Cox (1968 a) 4 0 ' Ehrlich, Miller & Walker (1970 b) ^ Cox (1976) 49)

° different with different atomizing fluid b two ranges of RH hastens decay rate c different for different organisms d survival descreases with increasing humidity

the shift and on the number of times the cells are subjected to a changing environment 43 \ Freeze-dried organisms undergo an " R H " shift when allowed to equilibrate to a higher environmental RH. Lyophilized cells of Serratia marcescens die rapidly at a water content similar to the equilibration to water vapour at an RH of approximately 90% 149 ' 15 °. 115>. Also, the decay rate of freeze-dried aerosolized S. marcescens increased at a high RH relative to the decay at low RH. However, a reverse relationship was found when these organisms were aerosolized from a liquid menstrua, e.g. heart infusion broth 5 ) . Similar unstable (low survival) zones have been found for dried Escherichia coli in vacuo at high RH 152) and for aerosolized E. coli at RH values of 100, 87, 85 and 50 % 28) . Cox 49) states that freeze-drying of bacteria to different water contents is analogous to aerosolization into atmospheres of different RH values (they have similar viability equations). Cox 6 | 7 ' 5 1 ) and Cox and Baldwin 1 U ) found that the survival of E. coli, after being aerosolized into a high RH environment, was sensitive to RH in a manner dependent upon the content of the spray fluid (Figs. 6 to 7). It has also been reported that the effect of an RH change before collection depended upon the strain of E. coli aerosolized and the nature of the spray and collecting fluids 7 ' 5 1 ) . However, aerosol clouds of Pasteurella tularensis, at low to middle range humidities (20—50%), showed a continuous biological decay rate independent of the type of collecting fluid used 38) . Generally, the behavior of aerosolized bacteria to RH differs with the dispersing fluid (Figs. 8, 9, 10) 6 ' 7 | 29 ' 5 2 A 5 \ sampling (assay) medium (Figs. 11 and 12) 6 - 40 ' 93 ', atmospheric components (Figs. 13 and 14) 6 ' 40 ' 46 », temperature 153) , and aerosol age (Fig. 15) 6 ' 101) .

J. E. Zajic, I. I. Inculet and P. Martin

66

Fig. 6. Survival of Escherichia coli (Jepp) in nitrogen at aerosol ages of a 0.3 s, b 3 s, c 2 min, and d 15 min. The solid dots are the results for bacteria sprayed from distilled water. The circles are the results for bacteria sprayed from 0.13M-rafiinose solution. The bacteria were collected in phosphate buffer. (Figure after Cox 7 ) )

Relative humidity (%)

Fig. 7. Survival of E. coli (Jepp) in nitrogen at aerosol ages of a 30 min, b 31.5 min, and c 3 h. The symbols are the same as in figure 6. Bacteria are collected in phosphate buffer for a and b. 1.0 M sucrose in phosphate buffer was used as the collection medium for b. (Figure after-Cox 7) )

6.2 Temperature T h e temperature of the environment plays a role in the stability o f airborne organisms. A n airborne particle in equilibrium with R H , at a temperature Tj is not in the same condition as o n e in equilibrium with R H at T 2 8 0 ) because o f the difference in the activation entropy. T h e activation energy associated with the death mechanism

67

Basic Concepts in Microbial Aerosols

Fig. 8. Survival in air of Pasteurella tularensis LVS. Symbols: o disseminated from the culture medium; • disseminated by air in a freezedried state (powder); A disseminated from distilled water. (Figure after Cox and Goldberg 45 ') 80

60

40

Relative humidity

20 (%)

Fig. 9. The effect of relative humidity on the death rate of Serratia marcescens aerosolized from various solutions (see symbols on figure). (Figure after Webb 22) ) 40

60

Relative humidity

(%)

and the activation entropy for the decay of the bacterial aerosol can be calculated by plotting the log k (death rate) against the reciprocal of the absolute temperature to yield an Arrhenius plot. Webb 80) showed from this plot that the decay rate (both in the 0— 1 s and the 1 —10 s sampling intervals) appeared to follow first order kinetics. For each 10 °C increment of temperature the death rate of Micrococcus Candida was found to increase 2 to 3 fold 9 5 ' 1 5 4 ) . This is similar to the Q 10 for most biological processes. Other results

68

J. E. Zajic, I. I. Inculet and P. Martin

Fig. 10. The aerosol survival of E. coli B sprayed from distilled water into helium as a function of relative humidity. The solid dots are the results for bacteria collected into phosphate buffer. The x's are the results for bacteria collected into phosphate buffer + Msurcose. (Figure after Cox 40) )

show that, at a single RH value, an increase in temperature resulted in a decreased survival time. For example, the death rate of P. tularensis increased with an increasing environmental temperature from 24 °C to 35 °C 35) . Flavobacterium sp. 3 4 ) 5. marcescens, and E. coli34) show an increase in the aerosol death rate for a progressive increase in temperature from —18 °C to 49 "C. The latter death rates appear to be linear with temperature. At low temperatures (—40 °C to 3 °C), the death rate for various airborne microbes was either not noticeably influenced or decreased. At —40 °C, negligible death rates for aerosolized S. marcescens were reported and the effect of RH ranging from 20 to 80% was absent 155) . Below 0 °C, very low death rates have been reported for S. marcescens80) and Flavobacterium sp. 3 3 ) . However, reduced survival has been found for P. tularensis and for S. marcescens and E. coli35•34) at 40 °C. At high temperatures (49 °C), optimum death rates and minimal recovery viability for various organisms have been demonstrated. For example, Flavobacterium sp. at high temperatures ( > 3 2 °C), when compared to data for S. marcescens and E. coli, had approximately 1/3 the death rate of the other two vegetative organisms although this was a maximum death rate for this species 34) . At this temperature, thermal inactivation could occur which could lead to alterations in proteinaceous structures and enzymatic activities. In summary, the factors sensitive to the variation of temperature appear to include the physical characteristics of the slurry, the configuration of the particles upon dissemination 35) , temperature sensitive biochemical reactions, varying evaporation rates, the water response to temperature and RH, and structural modifications of bacterial components caused by temperature changes.

Basic Concepts in Microbial Aerosols

69

Fig. 11a—f. Survival of E. coli (Jepp) in nitrogen sprayed from 0.3 M-rafflnose at aerosol ages of a 0.3 s, b 3 s, c 2 min, d 15 min, e 30 min, and f 3 h. The solid dots are results for bacteria collected in phosphate buffer. The circles are the results for bacteria collected in M-sucrose. (Figure after Cox 7>)

6.3 Evaporation Evaporation is dependent on RH, temperature, and the composition of water surrounding airborne particles. Orr and Gordon 156) found a slight decrease in size of individual airborne bacteria with decreasing RH. For a range of bacteria, the viable decay of airborne cells occurred more rapidly in smaller particles than in larger ones 17) . A water droplet of 13 |im diameter, at room temperature, was calculated to evaporate to 0.9 |im in diameter in 0.4 s 2). Owing to the enhanced vapour pressure over an aerosol droplet surface, the water of an airborne droplet will evaporate even in saturated air through thermal distillation 6). Water movement out of the particle and intracellular environment may contribute to microbial aerosol death 2 ' 39) . The evaporation of water from an aerosol droplet results in a lowering of

70

J. E. Zajic, I. I. Inculet and P. Martin

Fig. 12. Aerosol survival versus time for Serratia marcescens disseminated into varying concentrations of oxygen. (Figure after Zentner 39>)

71

Basic Concepts in Microbial Aerosols

Serrotio marcescens

strain 8 UK

Staphylococcus epidermidis strain NCTC 7291

Fig. 14. The effect of relative humidity and aerosol cloud age upon the survival of various bacteria. The plots represent (reading downwards) survival of populations recovered from aerosols into impingers at 1.2 s, 5 and 30 min after generation. (Figure after Anderson et al. 60>)

its temperature. The change in temperature during the process of atomization will cause a change in the vapour pressure of the droplet surface. The chilling of cells of E. coli to 4 °C caused rapid death 1 4 8 ) . A large temperature drop could produce a similar response in an aerosol. Consequently, Webb 8 0 ) states that survival in air will be greater if the evaporation process is carried out at a higher temperature (i.e., the faster the evaporation rate the better the chance for survival). However, as noted previously, within a certain range, higher survival is attained at lower temperatures. Therefore, at a lower temperature, the temperature drop caused by aerosolization is minimized and the evaporation rate is slower than at higher temperatures.

72

J. E. Zajic, I. I. Inculet and P. Martin

RH could, at constant temperature and pressure, exert its influence by controlling the rate of evaporation from an aerosol droplet 9 6 , 5 0 ) . Even in a saturated water vapour atmosphere, the aerosol droplet, after being sheared off the tip of the atomizer by the airstream of the atomizer, will lose water 155) . The rate of evaporation can be determined by the particle size, the nature and composition of the dissolved substances in the droplet and the diffusion coefficient of the evaporating material. When the aerosol droplet is transferred from the liquid in the atomizer to a lower RH (air), evaporation occurs until a balance between vapour pressures is attained. Kethley et al. 1 5 5 ) estimate that this equilibrium will be attained in 0.01 to 0.1 s depending on the RH; the lower the RH, the more rapid is the evaporation. This could be an important reason for the high death rate reported in the first second of aerosolization. A rapid loss of water from cell particles in air could result in one or more of the following: (i) weakening and/or collapse of certain cellular structures; (ii) concentration of droplet or intracellular constituents to toxic levels resulting in cellular poisoning or osmotic lysis; or (iii) inactivation of cellular functions by an imbalance of water throughout the cells 43) . With aerosols of E. coli, Poon 96) concluded that the rate of death and the initial rate of evaporation followed a similar pattern with respect to temperature and RH. After free water in a droplet starts to evaporate, a point is reached at which cellular water begins to evaporate. This may cause protein-water bonds to break and may possibly start the inactivation process for airborne bacteria. If the water layer surrounding the microbe is evaporated within a very short time, the bonds of the water molecules will be broken almost immediately after the aerosol is sprayed 2). Evaporation seems to coincide with the phenomenon of increased survival with increasing RH. Hayakawa and Poon 2) found the death rate of airborne bacteria to be slower than the evaporation rate because of the different boundary thicknesses the water molecule has to cross in order to evaporate. The loss of water may result in the loss of protein-water bonds (protein stability) or the exposure of cell components to the oxidizing action of oxygen. Cox 4 0 ' 9 3 ) demonstrated that, with aerosols of E. coli, survival was greater at low rather than high RH in inert atmospheres. He concluded that the initial evaporation rates of the aerosol droplets did not influence the long term survival. Nevertheless, evaporation will proceed faster at low RH and this would seem to support the hypothesis of Webb 80) that the faster the evaporation, the greater is the chance for survival. Decreased recovery of Flavobacterium sp. from the aerosol at extremely high RH could be due, in part, to the equilibration of the airborne particles to a diameter larger than 5 |im 3 4 ) . However, equilibration of cells with the environment has been discounted by Cox 1 5 7 ' 2 9 ' as the cause of differences between survival of wet and dry disseminated aerosols. The importance of the evaporation rate of airborne cellcontaining particles has not been elucidated clearly but it would appear to play a part in the survival of microbial aerosols.

Basic Concepts in Microbial Aerosols

73

6.4 Cell Water Content and Osmotic Pressure

Water accounts for approximately 77% of vegetative cells 158) . Consequently, biological material can be said to be hydroscopic. A viable cell will always contain some moisture in the airborne state. The water within the cell, according to the ion resin model 159) , seems neither to be as tightly bound as, either clathrate hydrates 1 6 0 , 1 6 1 ) , or hexagonal ice 1621 nor as free to orient as liquid water. Cell water is somewhat situated between these extremes. Different studies suggest that the water content of bacteria is related to the RH of the environment 163>22-5-164). xhe rate of transfer of water vapour between the cell surface and the environment is also affected by the surrounding RH 5). Changes in the water content of lyophilized S. marcescens as a function of the aqueous vapour pressure was described by Bateman et al. 1 6 4 ) . Approximately 85 % of the original 'sorbed' water 150) was lost upon equilibration at 40% RH (20 °C) but 10 % or less was lost at 97 % RH. Hess 37) noted the major changes in water content of aerosolized S. marcescens (wet disseminated) to occur between 40 and 97% RH. Maximum aerosol stability was achieved at 40% RH- the maximum water loss value for lyophilized S. marcescens. It appears that the greater the water loss, the more stable is an aerosolized microbe. Cox and Goldberg 4 5 ' state that the control of aerosol survival appears to be through the water content of P. tularensis at the moment of aerosol generation rather than in the aerosol phase. This conclusion is based on their experimental results showing that freeze-drying and reconstituting the bacteria with distilled water before aerosol formation had little effect upon survival as a function of RH. Freeze-dried bacteria show a hysteresis-like retention of water that suggests that the stability of the bacteria may depend on the history of the particle and/or the direction of water transfer. Bateman et al. 1 6 4 ) explained that the hysteresis could result from processes of slow irreversible changes in the dehydrated organisms or from the existence of a potential energy barrier (e.g. super saturation, failure to dissolve, or the formation of crosslinked structures which do not expand reversibly on wetting). Aerosol survival, because of the similarities with freeze-drying viability, could rely on these processes. Cox 29) , in a study of wet and dry dissemination of P. tularensis, found differences in the oxygen toxicity for the two methods of aerosolization. This suggests that the bacterial water content and activity do not control aerosol survival. The diffusion rates of intracellular molecules and polymeric precursors is dependent upon RH in a zone(s) of water content similar to freeze-dried microbes, as previously implied. At RH values high enough to distend the hydrophilic cell structures to their maximum extent, survival with increasing RH should be considered in terms of the osmotic coefficient i> 164) . In aerosols, the osmotic pressure inside microbes is at an optimum level in a saturated atmosphere 6 '. Therefore, no net water should enter or leave. However, the loss of water from cell solution could result in the irreversible formation of hydrates which in turn could change the osmotic pressure.

74

J. E. Zajic, I. I. Inculet and P. Martin

Cox 6 ) states that as the RH decreases, the concentration of solutes in the evaporating droplets surrounding airborne bacteria increases and the osmotic lysis could occur from the loss of cell water.

6.5 The Direction of Water Flow The layer of non-living material surrounding airborne microorganisms is termed the immediate environment 165) . Puck 166) suggested that for a particle, in an atmosphere containing vapours which are completely miscible with water, condensation and solution will occur. The loss of intracellular water can result in an increase in the osmotic pressure. The cell can gain water in a high RH environment as a result of a decrease in the osmotic pressure as vapour is adsorbed. The concentration of materials which will be achieved in an airborne particle will depend upon the vapour pressure relationships and the nature and composition of the bacterial particle 155> . A change in cellular water (by rehydration or dehydration) can cause a physical disruption of cellular components or create an imbalance in metabolic activity 139) causing the death of the cell. The formation of hydrates within the cell structure depends on the direction of water flow across the cells' barriers 49) . Minimum survival for wet dissemination of P. tularensis is between 30 and 60 % R H 4 5 ) . Apparently, water leaves the bacteria as they come into equilibrium with the RH of their environment. For dry dissemination of these organisms, minimum survival occurred at 80% RH, i.e., these bacteria gain water as they come into equilibrium with their environmental RH 45) . Consequently, the direction of water flow is important with respect to the aerosol survival for differing RH ranges.

6.6 Dehydration The cause of death of aerosolized bacteria has been attributed to the effects of dehydration 177,99) . The primary lethal effect could be intramolecular rearrangement, concentration of toxic products, or an imbalance of metabolic functions 39) . Observations on the variation of airborne bacterial death rate with variations in the RH led Dunklin and Puck 811 to suggest that the degree of cell dehydration is directly related to the sensitivity of aerosolized bacteria. Webb 80) attributed the lethal effect of dehydration of a structural change in an essential macromolecule when bound water is removed in the aerosol state. Neither strong salt solutions nor a high concentration of glycerol caused a larger death rate than that evinced in the airborne state 155) . However, comparisons of the kinetics of viability loss of aerosolized and freeze-dried bacteria and viruses show a similarity in their results 36> . Dehydration, therefore, may not be the sole cause of death of aerosolized bacteria but just a component of the overall death mechanism.

6.7 Rehydration Rehydration of bacteria seems to be involved in the RH influence on aerosol survival 6 - 7 ' 5 1 ' 4 0 ' 9 3 - 9 8 - 1 5 7 - 2 9 - 1 1 1 ' 3 1 ' 5 2 - 1 6 7 ' U 5 - 4 3 - 4 5 ) . Rehydration has been associated in the death mechanisms of dried E. coli aerosolized into a RH above 70% m - 5 2 ) .

Basic Concepts in Microbial Aerosols

75

In nitrogen atmospheres, the death mechanism of E. coli seems to be influenced by the way water reenters the bacteria during (Cox 98) ) and after collection 40) . Cox 6 ' 7 ) attributed rehydration as the cause of death of E. coli B and E. commune sprayed from raffinose solutions into nitrogen at high RH. The same mechanism could be involved at high RH in atmospheres of argon and helium, which were not found to be completely biologically inert 40) . Lysis on rehydration has been found not to be the primary death mechanism of microorganisms in aerosols 29) . The hysteresis effect can explain the differences in survival vs. RH results for wet and dry disseminated bacteria 49 '. However, one component involved in the differences encountered between survival of wet and dry disseminated aerosols is assumed to be rehydration 157,29) . The process of dehydration or rehydration in an aerosol seems to be involved with the formation of hydrates and the bonding of water in the cell.

6.8 Hydration and Adsorption The surface tension of moist particles governs the mole fraction of adsorbed vapours. Lowering of the surface tension of water by certain chemicals results in selective adsorption of these substances onto the surface and an increase in the mass of the particle. An example of some of these substances are proteins and carbohydrates. Water soluble vapourized organic compounds with a surface tension less than the bacterial droplet surface tension would also be selectively disolved The response to relative humidity shown by airborne microbes could be a result of the formation of hydrates of key biological molecules 1 6 9 , 1 7 0 , 1 6 8 ) . The hydrates could be formed in a semireversible or functionally irreversible manner. Couper et al. 1 7 1 ) have shown that enzyme inactivation can occur at certain RH values. The inactivation could be a result of the formation of different hydrates, some of which may be toxic or irreversible. Cox 6> hypothesized that a biologically inactive moiety may be produced from hydration of a DNA molecule in a semi-reversible manner by a process of aerosolization. However, Benbough 52) has concluded that DNA synthesis is not impaired by aerosolization of E. coli. Phage T7 in aerosols of E. coli collected at high RH were able to r e p r o d u c e m ) , which rules out impairment of DNA synthesis. Alterations in RNA synthesis did not seem to be adequate enough to drastically suppress colony formation and T7 production 40) . Cox et al. 3 6 ) and Cox 49) have derived viability equations that are based on the hydration of subpopulations of a hypothetical molecular species. This biochemical moiety is proposed to form a series of hydrates when exposed to an environment of low water activity. Protein (enzyme) hydration could be the site of the RH effect on survival.

6.9 Water Bonds and Proteins Damadian 159) has proposed the theory that cell water orientation is important in the transport system of microorganisms. Nuclear magnetic resonance studies have

76

J. E. Zajic, I. I. Inculet and P. Martin

demonstrated that cell water is more structured than elemental water and less structured than ice 172) . Structured water advocates maintain that structured cell water interacts with charged macromolecules inside cells. Damadian 100) proposes that a cellular resin is present in the cell and is composed of a fabric of charge groups of the structural polymers of the cell (nucleic acids, proteins, and phospholipids). Water molecules act as dipoles and line up against the charged groups (side chain amino and carboxyl groups and phospholipid and nucleic acid phosphates) of these structural polymers. The result is concentric layers of oriented water molecules about a macromolecule. As the distance from a charged macromolecule increases, order in the water layer decreases. Water structuring, because of its dielectric constant (which influences the force of attraction between two ions), regulates the selection of one counterion over another 173) . The removal of the most firmly held water molecules results in some loss of bacterial stability 163) . Ferry et al. 95) suggest that it is the rate of transmission of water through the cell boundary that is involved in death in bacterial aerosols. This movement of water molecules in and out of the cells in an equilibrium system could result in a collapse of the natural structures of the cell protein. For example, cells of E. coli that are freely permeable to ions, small molecules, sugars and peptides are air sensitive. Cells of Staphylococcus citreus and S. aureus, that are not so permeable, are aerosol stable 4) . Consequently, it can be stated that, if water is semi-structured with a wall or membrane component, the more loosely knit or permeable the cell barriers is, the easier it is to lose or gain cell water and the more unstable the cell is in the aerosol. With air-dried organisms in aerosols below 80% RH, death due to toxic chemicals, irradiation or dessication can be a result of removal or reorientation of bound water 174> 175) . The L-amino acids that compose peptide chains are the primary structures of protein molecules 176) . Protein groups have differences in their affinity for water which would allow for a large variation, from weak to strong, in the strengths of the protein-water bonds 8 0 ) . In an airborne cell, bonded water exchange with atmospheric water vapour could be determined by the strength of protein-water bonds. The many weak bonds could cause cell death but it might take breakage of only a few strong bonds to cause the same result 80) . An increased exchange of water molecules with the environment increases the possibility of a hypothetical irreversible change in the protein structure as a result of the removal of bound water. Webb 80) postulated that a dynamic system exists in the aerosol similar to the one shown below: PROTEIN • [H 2 0] n + H 2 0(g) PROTEIN • [H 2 0] n + X + ENERGY PROTEIN • [H 2 0] p + (H 2 0]j If the rates of water exchange between water vapour and protein bonded water molecules governs the rate of aerosol death, the relative strengths of the proteinwater bonds can be indicated by the inactivation energy for death 80). Small values of E would suggest that bonds of low energy level are involved in the death mechanism. Ionic, hydrogen, or phosphate bonds may take part in this interaction. Webb 80) found that, as the cell ages in air, more energy is required to produce its death (i.e. increasing activation energy (E) for death with increasing aerosol age).

Basic Concepts in Microbial Aerosols

77

If energy is available, the death rate during the latter stages of aerosolization increases more rapidly than that during the earlier stages. The difference between the E values for inactivation of wet and dry cells at low temperature could be due to the relative amounts of water in the cells. If there were several water layers covering one site of the protein water bond, an 'attack' on this vital site would be more difficult than if no protective layer existed 80). Bound water appears to occupy strategic positions in protein molecules and only certain chemicals forming hydrogen bonds analogous to protein-water bonds can replace them and maintain the biological integrity of the macromolecule 2 2 , 1 7 7 ) . The position of the bond in the molecule also governs the lethality of a protein-water bond rupture or formation. Cox and Baldwin 178) found that the gaseous atmosphere is involved in the death processes which occur at high RH, independent of oxygen. Gas hydrates which are able to modify a water lattice can be formed by nitrogen and argon. These two gases and helium can confer stability to a more polarizable solute clathrate that is present 179) . Therefore, the slight lethality of atmospheres of nitrogen, argon and helium may act through the modification of cell water structure. If a biochemical moiety (A • nH 2 0) is present in airborne bacteria, most of the water in this solution will evaporate between 60 and 100 % RH 36) . Below 60 % RH, sorption isotherms for bacteria show that water losses are small 1 8 0 ' 1 6 4 ) . An environment conducive to the existence of species A • nH z O has been hypothesized to occur in the interspace between the cytoplasmic membrane and the cell wall 36) where the toxic action of oxygen is thought to occur m . 4 °. 9 3 . 5 3 )_ Detachment of bacterial components involved in the transfer of substances through the bacterial membrane has been shown for two strains of aersolized E. coli. Incubation of the bacteria, after collection, with leaked components, partially restored transport activity of methyl-(a-D-gluco)pyranoside 62) . The leaked components were not identified but they might be counterions mentioned in the cellular resin model 1 7 3 ) or constituents of an active transport mechanism.

7 The Stability of Mycoplasmas, L-Forms and Algae in Air Bacteria and mycoplasmas differ greatly in cell size, composition and morphology of the cell boundary. Nevertheless, the uniqueness of these microbes does not afford them any less stability in air within a certain relative humidity (RH) range 4 1 , 4 3 ) than most other bacterial species 50). Airborne M. pneumoniae equilibrated at a dry or wet humidity range was shown to be highly sensitive to the effects of atmospheric moisture immediately after an abrupt change to a mid-range humidity 43) . A sudden change in RH which occurs in natural atmospheres can modify the biological stability of these microorganisms. The survival of this organism was also found to be a function of both R t i and temperature but the effects of temperature could only be observed if some water vapour were present 32) . Aerosolized streptococcal L-forms respond to RH in a manner similar to mycoplasma 4 1 , 4 3 ' 4 2 ) . Since mid-range RH values were also the most lethal for certain airborne bacteria 81 • 144 ' 60) , it seems that the microbial death mechanisms

J. E. Zajic, I. I. Inculet and P. Martin

78

involved in bacterial aerosols is independent of unique morphological features associated with the cell wall. Atomized algae (Nannochloris atomus and Synechococcus sp.) evinced the greatest loss in viability during the first minute after atomization. This initial inactivation is similar to other aerosolized microorganisms 181 • 145) . The subsequent long term survival was a function of RH 48) . The blue green algae Synechococcus sp. showed a pattern of survival resembling that of bacteria in that it remained viable throughout the RH spectrum. The eukaryote, N. atomus, remained viable only at near saturated humidities 48) .

8 The Effects of Radiation on Airborne Microbes Disinfection of airborne microbes by ultraviolet (UV) irradiation is dependent upon certain conditions including type of microorganism, suspending menstruum, atmospheric humidity, volume of space, quality of the irradiation, strength and length of the UV ray, total exposure, uniformity of exposure, and air motion. Ultraviolet irradiation killing of airborne microorganisms has not been found to depend on any RH effect 1 8 3 , 1 8 4 ) . However, Gates 184) and Riley and Kaufman 4 4 ) Table 6. The effect of simulated solar radiation, type of collection medium (agar) aerosol age, and tracer on the aerosol survival of Serratia marcescens. (Data after Dorsey et al. 5 5 ) Sampling time (min)

Light intensity Cmcal' cm 2 m i n - 1 )

Per cent recovery with indicated medium 8 Blood agar base Uranine

Casitone agar No uranine

Uranine

No uranine

b

53.9 56.6 49.2 41.8 33.2 16.9

(40) (39) (41) (36) (35) (23)

46.4 45.4 41.2 30.5 22.2 12.0

49.6 51.9 44.6 36.8 29.6 14.9

(37) (39) (36) (37) (35) (24)

(41) (29) (31) (26) (35) (36)

44.4 (23) 38.6 (31) 31.3 (40) 15.5 (26) 3.1 (49) 0.03 (116)

40.6 38.3 26.9 17.2 6.9 0.4

(36) (33) (30) (18) (36) (50)

30.5 (36) 24.4 (47) 15.4 (47) 2.5 (46) 0.08 (89) 0.00003 (175)

46.9 (44) 34.3 (22) 23.5 (20) 10.0 (41) 0.9 (45) 0.007 (69)

2 4 8 16 32 64

0

2 4 8 16 32 64

32

45.7 (27) 40.8 (32) 34.0 (35) 18.5 (30) 3.6 (44) 0.05 (89)

41.7 42.2 30.9 21.3 8.4 0.4

2 4 8 16 32 64

70

32.6 (35) 27.4 (42) 18.1 (48) 3.6 (41) 0.1 (80) 0.00006 (160)

49.6 (41) 37.5 (26) 25.5 (23) 11.6 (35) 1.0 (44) 0.009 (63)

a b

48.9 48.8 44.5 34.9 24.9 14.1

(46) (41) (40) (45) (54) (51)

(40) (37) (37) (44) (52) (50)

Mean of six runs for each condition Figures in parentheses are coefficients of variation rounded off to the nearest whole number

Basic Concepts in Microbial Aerosols

79

demonstrated a loss of UV sensitivity above R H values of 60 %. Response variations between organisms has also occurred. For example, aerosols of S. marcescens in the middle RH range were found to be more sensitive to inactivation 441 than other organisms 24>. Kundsin 26) found that droplet nuclei of bacteria and mycoplasmas were destroyed instantly by UV irradiation with very few survivors. The same study demonstrated that L-forms and Candida albicans retained 11 to 31% of their droplet nuclei after irradiation. Protection was hypothesized to occur through the action of budding or clumping. Simulated solar (xenon) radiation caused a significant decrease in viability with aerosols of E. coli or S. marcescens; Table 6 55) . Their decay curves showed at least two mechanisms of inactivation, one due to aerosolization, the other to irradiation. When sodium fluorescein was used for the assay method, an additional adverse effect was found. The adsorption of the radiation by nucleic acids can form thymine polymers which could be additive to the lethal effects of aerosolization.

9 Protecting Agents in Aerosols Hambleton 9 4 ' found some substances that improved the survival of bacteria in an aerosol. The mechanism(s) of protection could be the limitation of water loss from the microbe 6 ) , the replacement of structural water 22) , or the interaction of the substance with specific intracellular sites 52) . Cox 6) hypothesized that good protective agents for E. coli aerosols must prevent or moderate both the toxic action of air and the rate of viability loss within critical relative humidity (RH) ranges. Polyhydroxy compounds have been used as protective agents for aerosolized bacteria 22) because they form supersaturated viscous layers around bacteria and limit the loss of water 6 ' in the same way as those bacteria that possess a high lipopolysaccharide content 5 2 ) . Micromolar quantities of paramagnetic molecules such as M n + 2 , N O ^ , I " , and compounds such as ascorbic acid, aminothiols, and reduced dyes have been used and afford protection through a reaction or stabilizing action on the oxygen inactivating site 52) . Protective agents (spray additives) differ in their ability to penetrate the cell wall 6 , 5 1 '. Zimmerman 1 8 " suggested that nonpermeable sugars protect airborne microorganisms through a plasmolytic dehydration of the microbes. Hambleton 94) stated that rafflnose and dextran do not penetrate the cell wall, glucose and sodium glutamate (which does not permeate the bacterial membrane) entered very slowly, and glycerol was capable of penetrating the bacteria rapidly. Effective concentrations of agents that form supersaturated solutions around bacteria 6 ', aid survival of aerosolized microbes by preventing changes occurring in the bacterial wall and membrane and by protecting them from the lethal effects of extracellular enzymes. Protective agents do not operate through a modification of the initial evaporation rate of the aerosol particle 6 , 4 °'. Slight protection of E. coli K 12 Hfrl aerosols at high R H was found for glycerol and rafflnose, at low R H glycerol was toxic but rafflnose was highly protective 93) . The positioning of rafflnose outside the cell wall was able to confer stability in a similar manner to that for E. coli Jepp51Cox49) states that the protective action of

80

J. E. Zajic, I. I. Inculet and P. Martin

raffinose is related to its stabilizing action upon hydrates of a hypothetical active biochemical moiety. Cox 2 9 ) found that peptone broth spraying fluid either interacts with oxygen to yield a toxic product or does not completely protect against oxygen-induced death since survival in air for Pasteurella tularensis was less than in nitrogen. When a 1 % peptone diluent was used in the viability assay medium, the survival of P. pestis aerosolized into a high RH (65 and 87%) was adversely affected. However, heart infusion broth, used as diluent, increased the number of viable cells in aerosols held at high RH values 38) . S. marcescens recovery from the airborne state showed no change in either minimal or nutrient settling plate agar mediums. Plain gelatin liquid impingement medium, however, showed evidence of enhanced recovery with the highest recovery resulting from the enrichment of the medium 186) .

10 Serratia marcescens Serratia marcescens has been used by Orr and Gordon 156), Dimmick 5), and Cox et al. 3 6 ) in aerosol viability studies. The microbe is relatively nonpathogenic and easily grown in batch conditions. The distinctive red pigmentation allows for quick identification on agar plates in the presence of contaminating microbes. S. marcescens is a facultative anaerobe widely distributed in water, soil, and food. It is occasionally found in pathological specimens and has been identified in causes of outbreaks of nosocomial infections 187 ' 188) . Bacteriocins (marcescins) produced by this microbe, are proteinaceous antibiotics that could play a role in noscomial infections 189) . The synthesis of a prodigiosene (prodigiosin, Fig. 15) causes a red pigmentation in colonies. Maximum prodigiosin production has been attained with 96 h (senescent) cultures at 27 °C 190) . Prodigiosin synthesis is a variable characteristic (as shown by the appearance of white colonies) influenced by culture conditions. Nonproliferating (stationary phase) cells synthesize this pigment under alkaline conditions with amino acids and acetate as the substrates. Inhibition of prodigiosin production is

Ch^tCH 2)301^ MAP

CH2[CH2)3CH3

MBC

\

K |

'N' H

/

,

\

/

A N H

HN} — (

OCH3 ^ — ( f

Pvxx ,

„

„ CHO

JK A J N V^ N

CH 3

Prodigiosin

Fig. 15. The structure of prodigiosin and its precursors. MAP denotes 2-methyl-3-amylpyrrole; MBC denotes 4-methoxy-2,2'-bipyrrole-5-carboxyaldehyde. (Figure after Williams l90) )

Basic Concepts in Microbial Aerosols

81

brought about by the addition of glucose, phosphate, or chloramphenicol to the medium. Prodigiosin has been categorized as a secondary metabolite of S. marcescens 190). The monopyrrole precursor (2-methyl-3-amylpyrrole) has been shown to require at least four enzymes for its formation and five enzymes have been identified in the biosynthesis of the bipyrrole (4-methoxy-2-2'bipyrrole-5-carboxy-aldehyde); a tenth enzyme is involved in the coupling of these two components to form prodigiosin. The pigment is located in the cell envelope 190). Aerosolization of S. marcescens into a standard test condition (20 °C and 65 % RH) showed a variation in response that was not completely explained by the appearance of coloured variants (differing prodigiosin or colour concentration) from the blood-red wild type parent strain 186) . However, the reduced airborne viability was correlated to the emergence of a wide variation in the morphology and dimensions of the cells. An isolated variant culture (white, pink or red in appearance) was noted to decrease in airborne stability with an increase in the variety of cellular forms and sizes 186). Cell envelope glycoproteins, extracted by sodium dodecyl sulphate, from isolated walls of two strains of S. marcescens were found to be very similar in chemical composition 191 '. Both fractions contained approximately 50% proteins and 10% carbohydrates. The only sugar components identified in the carbohydrate moiety were glucose and glucosamine (refer to Fig. 4 for peptidoglycan structure). Orr and Gordon 156) , using the Millikan experiment, calculated the density (w/v) of aerosolized S. marcescens cells. The values of the density were found to be similar, over RH values from 40 to 100 %, to liquid-phase sedimentation studies which indicated a density from 1.057 to 1.062gem - 3 . At RH values below 40%, calculations showed a sharp decrease in the apparent density to a minimum of approximately 0.913 g e m - 3 at the lowest RH studied. The decrease in density was attributed to the inclusion of air into spaces produced by a wrinkling of the cell wall at low RH values. The average diameter of S. marcescens cells, aerosolized from beef extract broth at 20% RH, was found to be 0.62 |im ( + 0.02 standard error) when measured by an electron microscope after collection on an oil film155). Orr and Gordon 1 5 6 ) measured the radius of S. marcescens cells at the same RH but aerosolized from distilled deionized water by means of electron photomicrographs and found a mean radius of 0.4 |im (or 0.8 |im diameter). Evidently, the diameter of airborne cells depends on the spraying medium. Above RH values of 75 %, aerosols of S. marcescens have been found to be relatively stable with respect to viability in air 186) , in CO 4 6 ) , and after exposure to UV irradiation 44>. Below this RH value, viability variations in response to the aerosol state generally increased. Electron paramagnetic studies have shown that freeze-dried S. marcescens cells contain two unidentified molecular species that can react with oxygen 36) . Since the site of oxygen toxicity is believed to lie in the interspace between the cytoplasmic membrane and the cell wall 51) , these components are assumed to be located within this area. There are many stresses acting on the viability of aerosolized microbes which may act collectively or individually depending on the environmental conditions. If,

82

J.'E. Zajic, I. I. Inculet and P. Martin

indeed, there is a component of an airborne microorganism that is affected by aerosolization, as evidenced by the viability response of the cell, this component(s) must, at least, be located in the cell wall, have a site for oxidation, be neutralized in its toxic effect by chloramphenicol 31 \ and be susceptible to hydration or the absorbed water effect.

11 Review of Electrostatic Forces Relevant to Electrodeposition of Bacterial Aerosols Millikan in 1911 192) devised an experimental procedure to measure the charge unit on an oil drop in an electric field. In the absence of an electric field, a droplet falling through still air is acted upon by two forces (Fig. 16): a gravitational force Fg = m„g

(7)

where m p = mass of the particle, gm g = gravitational constant, c m s " 2 and an opposing frictional force of air drag F d = 6rairpvp

(8)

where r| = air viscosity, centipoise r p = radius of the particle, cm v p = speed of the particle, c m s " 1 . When the speed of the particle is constant (Fg = F d ) these forces balance at some downward velocity and the terminal velocity can be described by Stokes Law: Ppd 2 pg

where vt = terminal velocity, cm s _ 1 6 P = particle density, gm c m - 3 d p = particle diameter, cm.

Fig. 16. Forces that act on a falling particle in an electric field. F g is the gravitational force downward: F d is the drag force upward; F e is the electrical force in either direction depending on the polarity of the particle, sign of the electric field, and surface charge on the particle

83

Basic Concepts in Microbial Aerosols

If the particle is aerosolized into a uniform electrical field, the total force acting on the particle is a combination of Eqs. (1) and (2) and the electrical force: F

«down =

F

8 +

F

el ~

F

d

= m p g + qpE — 6jtrpvpT)

(10)

where F e l = the electrical force exerted in the downward direction, dyne. With no acceleration, the terminal velocity can be expressed as: _ VpgQp + q P s

(1))

67ir p r|

where Vp = volume of the particle, c m - 3 . However, aerosol particles can accumulate charge in the process of being aerosolized (e.g. contact electrification). The aerosol particle density in a given space decreases as a result of charge repulsion which causes the particles to migrate to the boundaries of the enclosed space. This decrease in particle numbers can be expressed as 1 »: 2L = nii

where n ^ t qp E Kp

= = = = = =

! no , 1+ qpKpt

(12)

cell concentration at time t, particles • c m - 3 initial cell concentration, particles • c m - 3 time, s charge on particle, coul permittivity, coul2 d y n e - 1 c m - 2 particle mobility, cm V" 1 s" 1 .

Cunningham's correction factor accounts for the discontinuous nature of air on a small size scale (near the mean free path of molecules in air). For very small particles, (S, O, x, t) - n T ] X(x, t)

(95)

and H(S, O, x, t) = n m

S(x, t) 0(x, t) ; : ' ' K s + S(x, t) K 0 + 0(x, t)

(96)

where u x is the specific death rate and x = x or x*. Cell growth rate, R x , is linked to substrate uptakp rate, R s , by the substrate yield coefficient, Y x/S (t) and to the oxygen uptake rate, PLQ, by the oxygen yield coefficient, Y x/ o(t). For more details on these models and others (fed batch, extended culture, continuous models) and for the numerical solution of the necessary differential equation systems, for parameter identification and process simulation carried out on a hybrid computer, see Refs. 1 8 1 • 269

6 Experimental Systems 6.1 Apparatus Four different tower loop reactor types were used for the investigations: Two concurrent air lift tower loop systems (a single-stage and a ten-stage reactor) and two counter current tower loop systems (a single-stage and a three-stage reactor with different bubbling layer heights.

124

AIR

K. Schügerl

IN

250 SHAKE

501 concurrent reactor,

main

bubble culture

columnloop

ml FLASK

first subculture

toi STIRRED

TANK

second subculture

a Fig. 4 a. Schematical view of the single-stage concurrent (air lift) tower loop reactor (Reactor 16,153,154) A water supply for steam generator; B flow meter; C pressure reducing valve; D pressure gauge; E gas flow meter; F valve; G filter for air sterilization; H three-way-valve ; J condensation collector; K sampling; L alkali reservoir; M substrate reservoir; N 0 2 -electrode; O pH-electrode; P pump; R exhaust gas cooler; S recorder; T engine for mechanical foam destroyer; U porous plate; V amplifier; W heat exchanger; 0 2 0 2 gas analyzer; C 0 2 C 0 2 gas analyzer; pH pH control

Characterization and Performance of Single- and Multistage Tower Reactors

125

Fig. 4b. Schematical view of the ten-stage concurrent (air lift) tower loop reactor (Reactor g j 17.178,305) PC pressure-reducing valve; M l , M2, M4 pressure gauge; SI, S2 gas flow meter; VI, V6, V7 throttle valves, V2, V3, V9, V10 three/two-way valves; F1 sterile filter; SM mechanical foam destroyer; F2 heated exit gas section; Kl—K3 condenser trap; PI membrane compressor; V4, V5, V8 shut-off valves; P2 centrifugal pump; N level control and overflow; D inductive flow meter; 0 2 L dissolved oxygen measuring electrodes and amplifiers; pH pH meter and control; B supply tank; T temperature meter (Pt 100); M3 difference pressure gauges; 0 2 paramagnetic 0 2 analyser; C 0 2 infrared C 0 2 analyser; DTU data transfer and storage unit; (1) exit gas; (2) condensed water; (3) sampling; (4) feed; (5) air in; (6) steam; (7) medium exit; (8) cooling water; (9) electrical heating; (10) drain. The instruments for bubble size and turbulence measurements are not given in this figure

126

K. Schiigerl

6.1.1 Single- and Ten-stage Concurrent Air Lift Tower Loop Reactors A stainless steel (V4-A) single stage bubble column (air lift) loop reactor, 15 cm in diameter, with a 275 cm high bubbling layer and a stainless steel porous plate (17.5 |im mean pore diameter) was employed for the cultivation of Hansenula polymorpha and Escherichia coli (Fig. 4). This tower will be termed Reactor A. Air and medium passed the column occurrently and the medium is recirculated through the 2.6 cm inner diameter loop. A mechanical foam destroyer was employed to make it possible to cultivate the microorganisms in the absence of antifoam agents. The apparatus was sterilized using water vapor (121 "C/2.2 bars). An oil free membrane compressor supplied the air, which was cleaned and sterilized by a filter system. The exhaust gas passed the mechanical foam destroyer, a cooler and a sterile filter, before it left the system. For more details see16».

The same tower was also used as a ten-stage reactor. Nine perforated plates were installed in the tower, which separated the tower into 10 sections of the following heights (from the top) (Fig. 4 a): 1 x 580 mm, 2 x 330 mm, 6 x 200 mm and 1 x 350 mm. Each of the perforated plates had 163 holes 3 mm in diameter and a relative free cross section of 6.53 %. E. coli was cultivated in this reactor, which will be termed Reactor B. For more details see Ref. 1 7 ) . 6.1.2 Single- and Three-stage Countercurrent Tower Loop Reactor The reactor consists of a 254 cm high stainless steel tower, 20 cm in diameter, which can be operated in a single- or three-stage mode. Fig. 5 shows the three-stage equipment. At the bottom of the column a perforated plate aerator, 12.4 cm in diameter, with 302 holes, each of them 0.5 mm in diameter, is installed. The tower is equipped with a mechanical foam destroyer at the head and a gas exit with a cooler. The main part of the liquid was pumped by a magnetic rotational pump from the bottom of the column through the 1.7 cm in diameter loop, back to the top of the column.

The single stage tower will be termed Reactor C. Separating trays were employed in the three-stage column at distances 66 and 116 cm from the aerator compartment. They consisted of a perforated plate with 540 holes, 0.5 mm in hole diameter (free cross sectional area 0.34%) and an overflow, 46 mm in diameter, with 20 and/or 40 cm overflow heights. The stages are numbered from the top to the bottom in direction of the liquid flow. The 40 cm overflow tower will be termed Reactor D-40; the 20 cm overflow tower Reactor D-20. An oil-free compressor supplied this equipment with air, which was sterilized by a sterile filter. The exhaust gas left the reactor through a cooler and a sterile trap. For more details see Ref. 87) .

Hansenula polymorpha was also cultivated in Reactors C and D.

6.2 Measuring Devices In order to characterize and control the reactors, several measurements were carried out on the systems and on samples taken from them. These measurements will be divided here into on-line and off-line measurements.

Characterization and Performance of Single- and Multistage Tower Reactors

127

Fig. 5. Schematical view of the three-stage countercurrent tower loop reactor (Reactor D) 8 7 , 8 8 ' 8 9 ) . 1 air supply; 2 N 2 supply; 3 three-way valve; 4 pressure reduction valve; 5 gas flow meter; 6 needle valve; 7 water vapor (121 °C); S.sterile filter; 9 valve; 10 condensation collector; 11 thermostatized water inlet; 12 sampling valve; 13 gas distributor; 14 resistance thermometer; 15 0 2 electrode; 16 pH electrode; 17 compartment separating tray; 18 overflow tube; 19 mechanical foam destroyer; 20 exhaust gas cooler; 21 cooling water; 22 waste water; 23 thermostatized water backflow; 24 inductive liquid flow meter; 25 alkali reservoir; 26 magnetic valve; 27 substrate reservoir; 28 pressure gauge; 29 heated exhaust gas tube sterile gas outlet; 30 safety valve; 31 exhaust gas; C0 2 C0 2 -analyzer (exhaust gas); 0 2 0 2 -analyzer (exhaust gas); T temperature measure and control; P liquid pump; V inductive flow meter

6.2.1 Reactors A and B On-line

measurements

— pH control (Ingold electrode, PD controller) — temperature control (resistance thermometer PT 100, PD controller)

K. Schügerl

128

— tension of dissolved oxygen at 13 positions along the column (by oxygen electrodes developed in the author's laboratory) — C0 2 concentration in the exhaust gas (Uras 2T Hartmann and Braun) — 0 2 concentration in the exhaust gas (Oxymat 2, Siemens) — alcohol concentration in the exhaust gas (flame ionisation detector, RS5 Ratfisch) — volumetric liquid flow rate through the loop (inductive flow meter, Krohne) — bubble size distribution (by means of 4 electrical conducting microprobes (see 4.3.2) — local gas hold-up (by means of 4 electrical conductivity micro probes (see 4.3.2) Additional

measurements

with E. coli:

— 2 constant temperature anemometers with 2 wedge shaped probes (DISA 55R32), to measure turbulence properties (see 4.3.6 to 4.3.11).

For more details see Ref. 273) . Off-line — — — — — — —

measurements

dry biomass (after separation by weight) cell concentration by optical density (PMQ2, Zeiss) glucose concentration (Polarimeter PM241, Perkin Elmer) ethanol concentration (gas Chromatograph, L400 Siemens) volume of feed substrate bubble size distribution (flash photography; see Ref. 226) ) by means of three windows gas flow rate (rotameter, Krohne)

6.2.2 Reactors C and D On-line — — — — — —

Off-line — — — — —

measurements

temperature control (resistance thermometer type PT 100, PD controller) pH control (Ingold electrode, PD controller) tension of dissolved oxygen in each of the stages and in the loop (again by oxygen electrodes) liquid flow rate in the loop (inductive flow meter, Krohne) 0 2 concentration in the exhaust gas (Oxygor 3, Maihak) C0 2 concentration in the exhaust gas (Unor 6, Maihak) measurements

dry biomass (after separation by weight) cell concentration by optical density (PMQ2, Zeiss) ethanol concentration (gas Chromatograph L400, Siemens) gas flow rate (rotameter, Krohne) volume of feed substrate.

6.3 Subcultures and Operations Hansenula

polymorphs.

In Table 7, the composition of the nutrient solution is given for the first (250 ml) and second (10 1) subcultures, as well as for the main cultures (32-50 1). For the glucose runs in the two subcultures and the main culture, 1 % glucose was used. For ethanol runs in the first subculture, 1 % glycerol in the second subculture, and 1 % ethanol in the main culture provided the energy source. The cultivations were carried out at 38 °C and pH 5 in batch, fed batch and in extended culture operations. For more details see Refs. 16 ' 87) . Escherichia coli The medium composition of the second subculture (10 1) and main culture (50 1) is given in Table 8. Cultivations were carried out at 28 and 30 °C, pH 6.7-6.9, in batch and continuous operation. For more details see Ref. 17) .

Characterization and Performance of Single- and Multistage Tower Reactors Table 7. Composition of the nutrient solution

16,153)

gl"1 (NH4) 2 so 4 K 2 HP0 4 • 3 H 2 0 NH2PO4 MgS0 4 • 7 H 2 0 NaCl KCl H3BO3 FeCl3 • 6 H 2 0 9'6

5 1 3 0.5 0.1 0.1 0.5 x IO"3 0.2 x IO"3

129

gl"1 MnS0 4 • H 2 0 ZnS0 4 • 5 H 2 0 CuS0 4 • 5 H 2 0 KI ( N ^ M ^ O a • 4 H20 Thiamine Biotin

0.4 x IO"3 0.4 x IO"3 0.04 x IO"3 0.1 x IO"3 0.2xl0"3 0.004 0.02 x IO"3

Table 8. Nutrient medium for cultivation of E. coli in stirred tank and tower loop reactors 17 ' 305) 1.0% glucose 1.0 % casein peptone 0.5 % meat extract 1.0 % yeast extract 0.25% NaCl pH 6.7-6.9 0.1-0.2% Desmophen 3600 as antifoam agent

6.4 Properties of the Employed Media Turbidity temperatures, TT, are mainly influenced by nutrient salts in the cultivation media. Since the nutrient salt composition and salt concentrations vary only slightly, T t is nearly constant during cultivation: T T = 53.3 °C during cultivation of H. polymorpha on substrate glucose and T t = 55.9 °C during cultivation of E. coli (For comparison see Table 6) Oxygen solubilities are given by Bunsen coefficients, a B aB =

abs. gas volume under stand, cond. dissohed liquid volume

The employed salts, glucose, proteins (casein peptone, yeast extract, meat extract) reduce the 0 2 solubility. However, since nutrient salt and dissolved protein concentrations vary only slightly during cultivation, the a B values are nearly constant: ocB ^ 22 x 10"3 during H. polymorpha cultivation on substrate glucose, at T = 38 °C, a„ ^ 23 x 1(T3 during E. coli cultivation at T = 28—30 °C. With decreasing glucose concentration (in a batch culture) a B increases 201 '

K. Schiigerl

130 For comparison: a B = 26.68 x 10~3 in water at T = 30 °C .

The viscosity of the medium was only slightly higher than that of water due to low cell mass concentrations used during cultivations. Therefore the diffusivity of 0 2 was also nearly the same as in pure water. The surface tension of cultivation media,CT,varies during measurement due to the denaturation of proteins on the gas/liquid interface. It takes a fairly long time to attain a constant value, which is called the equilibrium surface tension, a e q . With increasing cultivation time of Hansenula polymorpha, t, (in the absence of antifoam agents) creq diminishes from 58 mN m " 1 (Millinewton per meter), at t = 0 to 45 mN m _ 1 at t = 30 h 210) . The variation ofCTwith the measuring time, tM, can be described by a simple relationship 210) : log 2.3 (log V) = m log tM + log b where V =

~

CTeq

CT, — CTeq

CT isCT„at t = 0 CT isCT,at t and

m and b are constants which depend on operational conditions. In nonlimited growth of H. polymorpha, m diminishes with increasing tM from about 1.0 to 0.3—0.4 (in oxygen transfer limited growth) and/or 0.2—0.3 (in substrate limited growth), m is constant in limited growth operations. In nonlimited growth, the constant b increases with increasing tM from 3 x 10 - 4 to 0.2—0.4 (in oxygen transfer limited growth) or to 0.4—0.6 (in substrate limited growth) 210) . This variation can be explained by the change in the protein structure on the gas/liquid interface, which obviously influences the k L a values (see also 6.5.1.2). H. polymorpha cultivation media on substrate glucose have about the same coalescence promoting effect as pure water: m corr ~ 1.0. H. polymorpha media on substrate ethanol are, to a considerable degree, coalescence suppressing. However, because of strong foaming (no antifoam agent was used) it was only possible to determine m corr at low superficial gas velocities: m corr ~ 1.3 at wSG = 2 cm s" 1

217 218)

-

In the presence of antifoam agents, the coalescence promoting effect is strong. For H. polymorpha on substrate ethanol, m corr = 0.66 at wSG = 4 cm s _ 1

in the presence of silicon oil orDesmophen 3600 217.218>

E. coli cannot be cultivated in the absence of an antifoam agent because of uncontrollable foaming. In spite of the presence of antifoam agents, E. coli medium has a coalescence suppressing character: m

corr

=

1-4 at wSG = 4 cm s _ 1

in the presence of Desmophen 3600

Characterization and Performance of Single- and Multistage Tower Reactors

131

This property is responsible for the unexpectedly small Sauter diameters, d s , the high specific interfacial areas, a, and k L a values which were measured in E. coli media during cultivation (see also (7.2.1)). Foaminess is not only a function of protein and salt concentrations but also depends on the measuring time, t DG , which is necessary to attain the equilibrium surface tension, crcq (H. polymorpha): E = 4.23 x 104 (1 + 3.85 x 10 - 5 )-'dg where E is given in seconds; t DG in minutes. Thus, if t DG is small (e.g., 1500 min), £ is large (1000 s), and, conversely for high t D G (4500 min) X is low (2 s) 2 1 0 ) . Cell microflotation is influenced by several parameters. However, the most important factors are the foaminess and the maximum possible cell concentration in the foam. If the foaminess is small and the cell concentration in the medium, X p , is high, the cell concentration in the foam, X s , corresponds to the constant saturation value. If the foaminess is high and X p is low, X s is low as well, and the foam mainly carries aqueous solution. In batch operation and at a low cell concentration, X p , the first fraction has the highest (saturation) X s value. H. polymorpha can be microflotated directly from its medium, if the medium foaminess is adequate. This is the case in oxygen transfer limited growth. At low medium cell concentrations (X p ~ 2 g 1 _ 1 ) a high concentrating index (X s /X p > 20) and enrichment index (XS/XR > 500) can be attained 247 . Here, cR is the cell concentration in the rest liquid after cell flotation. Nonflocculating E. coli cells form a stable cell suspension. Nonflocculating yeast cells have very low sedimentation rates. Saccharomyces cerevisiae, for example, has a typical sedimentation rate of 0.276 cm h _ 1 , calculated from the displacement rate of the clear liquid/turbid liquid interface. The addition of CaCl 2 , Chitosan, H 3 P 0 4 , H 2 S 0 4 (pH 3), Dextrane, or DEAE cellulose has only a slight effect on the sedimentation rates. They vary within 0.2 and 0.28 cm h - 1 . The corresponding half lives, t 1/2 , which are necessary to reduce the cell concentration (extinction) to half of the original value, halfway up the cylinder, vary between 42 and 53 h. With 0.2 % CMC (Tylose C 300, Hoechst AG) the sedimentation rate increases to 2.11 c m h - 1 and t 1/2 diminishes to 11 h 256) . Flocculating strains of E. coli or S. uvarum have high sedimentation rates which cannot be measured using these batch methods but only by continuous methods. They are two to three orders of magnitude higher than the sedimentation rates of nonflocculating cells.

6.5 Properties of the Employed Two-phase Systems 6.5.1 Gas Hold-up, E G , Bubble Diameters, d and d s , Specific Surface Area, a, and Volumetric Mass Transfer Coefficient, k L a 6.5.1.1 Spacial Variation The measurements of the local properties of two phase systems indicate that radial profiles of 3 and d s in tower reactors are fairly uniform 3 6 , 1 7 8 - 2 2 6 . 2 2 7 . 2 7 3 - 2 7 4 ). Also

132

K . Schügerl

their longitudinal variations are fairly moderate except in the neighbourhood of the aerator 1 7 8 , 2 7 4 ) . The same holds true for the spacial variations of the local E G 's. According to Eq. (35), a is proportional to E G , (1 — E G ) _ 1 and d f 1 , thus the spacial dependence of E G is amplified in a. Especially at high superficial gas velocities, i.e. in the heterogeneous flow range, the radial profile of a has a shape of an error function, with its maximum in the column center 3 6 , 2 2 6 , 2 2 7 ) . Radial profiles of local k L a values have not yet been published. However, one can assume that they are similar to the radial profiles of a, since there is no reason to assume that k L is not constant in the tower cross section. The behaviour of these parameters near the aerator depends on the aerator itself and on the medium character. I f the primary bubble diameter, d p , is larger than the corresponding dynamic equilibrium bubble diameter, d e , or if they are equal, d and d s are also constant in the aerator vicinity. Only if d p -4 d e , does the bubble size increase with increasing distance from the aerator, x, due to coalescence, until d c is attained. How quickly d s ~ d c is attained depends on the medium property. F o r uniform longitudinal 3 and d s -profiles, the local gas hold-up, E G , is also constant and does not depend on the longitudinal position. This holds true also for the aerator vicinity. In systems with d p -4 d e , i.e. in which d s increases with increasing distance from the aerator, x, the local e G passes a maximum and then it diminishes with increasing x. In these systems a depends considerably on the longitudinal position, x, as can be seen from Fig. 6. a attains a constant value as soon as d s d e has been established. Also k L a exhibits a strong dependence on x in these systems in the vicinity of the aerator 8 0 " 8 1 ' 2 7 5 ) . Looking for the cause of this dependence, one can recall that in systems with d p -4 d e ) aerators have considerable influence on a, which depends on the medium significantly. In Fig. 7, a* values are plotted as a function of the superficial gas velocity with three different aerators (different d p 's), in coalescence suppressing medium. The large aerator effect diminishes, as a medium with slight coalescence suppressing effect is employed (Fig. 8).

8000

/ /

> S a

•

4000

•

® ©

/

jr

2000 /

A/

/X

/ / / / /

/

/

•

/

/

2

Wsg

cms"

-

Fig. 6. Specific interfacial area, a* = A'/V, as a function of the superficial gas velocity in a 258 cm higher tower reactor 14 cm in diameter by employing 1 % ethanol substrate and nutrient solution according to Table 7. Influence of the longitudinal position, x; 1 distance from the aerator, x = 41 c m ; 2 x = 115 cm; 3 x = 171 cm

Characterization and Performance of Single- and Multistage Tower Reactors

133

•

5000

3000

> ?

2000

©

N

/ /

i

/

•

/m

/

rfi

1000

Fig. 7. Specific interfacial area, a* = A'/V, as a function of the superficial gas velocity in a 258 cm high tower reactor 14 cm in diameter by employing 0.5% propanol solution. Influence of the aerator, x = 171 cm; 1 porous plate 5 urn in pore diameter; 2 porous plate 50 nm in pore diameter; 3 perforated plate 500 nm in hole diameter

4

2 WSG

1200 -O——

5

/

&

—

i

>4

M0

•S

2

4

cm s '

Fig. 8. Specific interfacial area, a* = A'/V, as a function of the superficial gas velocity in a 258 cm high tower reactor 14 cm in diameter by employing 0.5% methanol solution. Influence of the aerator. x = 171 cm; 1 porous plate 5 nm in pore diameter; 2 porous plate 50 Jim in pore diameter; 3 perforated plate 500 |im in hole diameter; 4 ejector nozzle 3 mm in diameter; 5 injector nozzle 4 mm in diameter

Wso

In coalescence promoting media, this effect is only slight (Fig. 9). This aerator effect on a is well documented in the literature 14 . 37 > 80 - 87 ' 277 > 278 >. This effect can be taken into account by assuming a variation of k L a with x near the aerator 16 ' 153,181 ' 279 >. However, the variation of k L a in the vicinity of the aerator is not only due to the space dependence of a, but also to that of k L 276) . Near the aerator the turbulence intensity is high, which causes a redistribution of the gas phase. With increasing superficial gas velocity, the specific power input increases which reduces d p 280) . During bubble redistribution, the bubble surface is renewed and a strong "deformation turbulence" due to considerable oscillation and shape variation prevails 281) . All of these phenomena increase

134

K. Schugerl

Fig. 9. Specific interfacial area, a* = A'/V, as a function of the superficial gas velocity in a 258 m high tower reactor 14 cm in diameter by employing distilled water. Influence of the aerator, x = 171 cm; 1 porous plate 5 nm in pore diameter; 2 porous plate 50 |im in pore diameter; 3 perforated plate 500 (am in hole diameter

kL. These spacial dependencies near the aerator of k L , a and k L a must be taken into account, if one desires to describe longitudinal concentration profiles of dissolved oxygen in the tower reactor.

In the tower loop reactors considered here, this spacial dependence of k L a was calculated by Eq. (97), 1 8 1 ' 2 8 2 ) : kLaE(t) exp - K S T ( t ) k L a(x,t) =

aL„

(97 a) 0 < x < oeU

E

(97 b)

^kLa (t) exp [-K S T (t)] aL R i x ^ L ,

where k L a E is the volumetric mass transfer coefficient at the gas entrance (x = 0) and K s t the coalescence factor The simulations of longitudinal concentration profiles of dissolved oxygen indicate that all profiles could be fitted by assuming a constant validity range of Eq. (97): a = 0.1 = z, where z = x/L R and L R = 276 cm, the height of the bubbling layer. Thus, in the tower reactors considered here, the variation of k L a is only taken into account in the range x = 0 to x = 27.6 cm. In this range, k L a diminishes exponentially with K ST according to Eq. (97a). In the range x = 27.6 to 276 cm, k L a is constant and given by Eq. (97b). This k L a is called k L a" here: kLa«(t) = k L a E ( t ) e x p [ - K s x ( t ) ]

(97c)

The reduction of k L a E at the reactor entrance to its spacial independent value, kLa", is characterized by the coalescence function, 4"*, 1 8 1 ' 2 8 2 ) : fit,) =

M'tti)

(98)

ME(ti)

which is considerably influenced by the medium properties and operation conditions. For example in the nonlimited growth range of Hansenula polymorpha cultivation

Characterization and Performance of Single- and Multistage Tower Reactors

135

0.7

/

0.6

/

0.5

/ /S/

0.4

0.3

Fig. 10. Mean relative gas hold up, E g , as a function of the superficial gas velocity in a 258 cm high tower reactor 14 cm in diameter by using porous plate 5 |*m in pore diameter. Influence of the medium properties, x = 171 cm; 1 distilled water; 2 0.5% methanol solution; 3 0.5% ethanol solution; 4 0.5% propanol solution

©1

0.2

0.1

©

/

y

U

2

4 WSG

© mm

•

yM

li

V

/ ®

©

2

4 Wsq —

Fig. 11. Sauter bubble diameter, d s , as a function of the superficial gas velocity in a 258 cm high tower reactor 14 cm in diameter by using porous plate 5 nm in pore diameter. Influence of the medium properties, x = 171 cm (For symbols see Fig. 10)

136

K. Schügerl

on substrate ethanol ^"(t) increases from 0.37 to 0.72 with the cultivation time, t. At growth transition from the nonlimited to the oxygen transfer limited state, ^"(t;) diminishes and then fluctuates around 0.50. When employing glucose as a substrate, ^"(t;) has a much lower value. It fluctuates around 0.16 during the first 6 h 1 8 1 , 2 8 2 ) . 6.5.1.2 Influence of Medium

Composition

The composition of the medium only influences E G , d, d s , a and k L a, if d p d e , or if d e is space dependent. In single-stage tower reactors d e is not space dependent, except in the aerator vicinity, thus in the predominant section of the tower, a medium effect is only expected if d p d e . Medium effects are well documented in these reactors 1 3 ' 3 7 - 8 0 " 8 7 ' 1 4 8 , 2 7 7 ' 2 7 8 1 . In Figs. 10, 11 and 12, E g , d s and a* are plotted as functions of the superficial gas velocity w SG , in a tower reactor with porous plate gas distributor (S urn mean pore diameter) employing coalescence promoting media (dist. water) and media with increasing coalescence suppressing effects (0.5 % methanol ethanol -» propanol solutions). However, it is difficult to find quantitative relationships between these parameters and the composition of cultivation media.

5000

3000

£>

f 1/ 1/

Ö

V

A

_

A —

2000

1000

'

W

SG"

cm s"

©

Fig. 12. Specific interfacial area, a* = A'/V, as a function of the superficial gas velocity in a 258 cm high tower reactor 14 cm in diameter by using porous plate 5 um in pore diameter. Influence of the medium properties, x - 171 cm (For symbols see Fig. 10)

Characterization and Performance of Single- and Multistage Tower Reactors

137

In Fig. 13, identified kLa values are plotted as a function of ethanol concentration. These kLa""s where evaluated from different cultivations of Hansenula polymorpha with ethanol substrate at the same superficial gas velocity (wSq = 1 . 9 c m s - 1 ) . Identified k L a values are also plotted for comparison. They were measured by Oels 1 3 , 2 8 3 ) in nutrient salt solutions at S = 0 and S = 5 g 1 _ 1 ethanol concentrations with the same aerator and at the same superficial gas velocity. With increasing ethanol concentrations, k L a" becomes larger. However, k L a" seems to depend also on the cultivation mode. At the same substrate concentration, the lowest kLa® values were found in oxygen transfer limited cultivations, probably due to higher dissolved protein concentration in the medium caused by higher cell death rate. The lower boundary of k L a" values for oxygen limited growth can be represented by a simple straight line. At S = 0 g I" 1 , k L a" = kLaS and at S D = 1.25 g l" 1 , k L a" = 2kLag (SD is defined as the substrate concentration at which kLa" attains 2kLajj). In nonlimited growth S D = 1 g l - 1 . For S ^ 3 g 1 _ 1 the dependence of m L on S can no longer be described by a linear relationship. This relationship holds true for antifoam free systems. Antifoam additives strongly influence k L a in tower reactors 1 4 , 2 1 7 , 2 1 8 , 2 8 4 ) . However, at low concentrations their concentration also plays a role. When employing antifoam control, the antifoam concentration in cultivation medium is usually unknown. At intermediate concentrations of antifoam additives, k L a does not depend on their concentrations. High concentrations of antifoam agents can cause foam formation and an increase in k L a. Because of the complex dependence of k L a on antifoam type, concentration and medium composition, no relationship is known to calculate this interdependence. However, when employing antifoam agents at intermediate concentrations, the calculation of k L a appears to be simple.

2000

1 M

• o -sc"1 1 5 0 0 •-•

1

e

k L a" ^

V

- k L a «o nation r a n g e with oiit cells lOels 1975)

— o i

a = 0.1

z

a

/

o

o

•

A AO O .

V

»

(

A

»

j

/ ••

i •

j

1000 lowest A A

k I i«

5 0 0 "- i1 OELS—< >

kLa 100 0

y

limit for all k°a values S< 6 g r

'

i i i i

- t -

.9 c m s -

i •

0

SD

2 S (gl"')

Fig. 13. Influence of ethanol concentration on the volumetric mass transfer coefficient, k L a, during cultivation of H. polymorpha 181, 271) . k L a" = 235 h " 1 , S D = 1.25 g r 1 . va batch cultivation • substrate limited cultivation o extended culture • oxygen transfer limited cultivation lowest limit for all k L a values T k L a variation range without cells 2 8 3 )

138

K. Schugerl

All cultivations investigated in the author's laboratory gave the same k L a value if an antifoam agent and its concentration were identical 2 1 8 ) . Since it is difficult to remove antifoam agents completely from the inner surfaces of bioreactors, investigations with them are quite tedious and time consuming, which is probably the reason why such investigations are scarce.

The few data published in the literature 2 I 7 , 2 1 8 ' seem to indicate that the addition of antifoam agents does not influence E G , but strongly reduces k L a. The reduction of k L can be determined at low w S G , where no coalescence occurs. At short bubbling layer heights, the diminution of k L a is mainly caused by the reduction of k L 2 1 7 ) . Only in higher columns does the increase in d s have to be taken into account. 6.5.1.3 Influence of Aeration

Rate

Aeration rate dependence of E G , d, d s , a and k L a in tower reactors is well documented in the literature (e.g. i3.3°-37,80-87,276-280) In tower reactors with an outer loop, two effects overlap: With increasing gas velocity, E G , a and k L a are enlarged in the homogeneous flow range similar to tower reactors ("tower effect"). However, in tower loop reactors increasing gas velocity also enlarges the liquid velocity which causes a decrease of the above mentioned parameters ("loop effect"). At low gas velocities, the "tower effect" dominates and at high gas velocities the "loop effect". Hence, with increasing gas velocity in tower loop reactors, k L a at first increases, often passes a maximum and then d i m i n i s h e s 1 6 , 1 1 • 1 4 8 , 1 5 5 ) . In Fig. 14 O T R is plotted as a function of aeration rate in a tower loop reactor during cultivation of Hansenula polymorpha. At 2 0 % saturation of dissolved oxygen concentration, O T R passes a flat maximum as a function of the aeration rate. This maximum occurs at a superficial gas velocity of w SG = 2.16 cm s - 1 . The agreement between the superficial gas velocities at this maximum of O T R found in the author's laboratory 17 ' 155 > and determined by Weiland 1 4 8 ) is satisfactory.

vvm

Characterization and Performance of Single- and Multistage Tower Reactors

139

1 o O

i 1

O O

•7

!

/

/

o

^y y

o

y

/

•V /

0?

-

N

S -

O F

" ^ - D A T A O E L S

• X

'

y

y* Ck/

A

y

-

Fig. 15. Mass transfer coefficient ratio, m L , as a function of the substrate (ethanol) concentration during the cultivation of H. 181 282 polymorphe - > 1 nonlimited and substrate limited growth 2 lowest limit for all k L a values ( 0 2 limited growth) I m L variation range without cells 2 8 3 ) S D = 1.25 g l " 1 , U Q Q = 1.9 cm s " 1

imi t e d

*

' N1

no n

l i m i ! e d

l i m i t e d

2

3

g

I"

S —-

At low superficial gas velocities and liquid circulation rates the "tower effect" dominates. In this range the influence of w| G on kLa® can be taken into account by Eq. (99), by separating the substrate effect from the gas velocity effect l 8 1 ' 2 8 2 ) : k L a°(w| U! S) = k L a£(w SG ) m L (S)

(99)

where m L (S) is a coalescence factor (—). Since common nutrient salt solutions have similar coalescence suppressing effects, this effect was integrated into kLa£ of Eq. (99). With m L (S) the coalescence active substrate concentration was taken into account. Fig. 15 shows m L as a function of S for different cultivations. The relationship between m L and S can be described by Eq. (100): m L (S)

(100)

where m

s

L (-)

(gr 1 )

S D = 1.25 for oxygen transfer limited growth (curve 2) S ^ 6 gl" 1 Su = 1.0 for nonlimited growth (curve 1) S g 4 g l - 1 In Fig. 16, kLa,j is plotted as a function of the superficial gas velocity at the reactor entrance, wf G , for cultivations with ethanol and glucose as well as for nutrient salt solutions. This figure indicates again that a low w| G values, the agreement between kLaJ5 in the tower (model media) and in the tower loop (cultivation media) are satisfactory. However, with increasing gas velocity, the "loop effect"

140

K. Schugerl 8 00

h"1

t

¿00 Fig. 16. Influence of the superficial gas velocity, UQO, on the volumetric mass transfer coefficient in the absence of ethanol during the cultivation of H. polymorpha 181 •282'. o substrate limited growth, ethanol substrate, S = 0 g r 1 o glucose substrate S =j= 0 g l" 1 — range for 0.5 % nutrient salt solution in tower reactor 2831

200

0

0

1

2

3

U

cm S"1 6

increases and at high wf 0 it dominates. Thus with increasing w| G , kLag passes a maximum between 2 and 3 cm s - 1 , in contrast to tower reactors in which no such maximum exists. There is a fairly large scattering of the kLag data due to the complexity of the biological media. However, there is a clear tendency for w| G 5s 3 cm s" 1 i.e. kLao increases with w| G . This can be approximated by Eq. (101): kLaS = 133.3 wf G ;

wSG g 3 cm s" 1

(101)

where kLaS ( h - 1 ) and w| G (cm s" 1 ). For wfc > 3 c m s " 1 , kLa£ diminishes with increasing wf G . Relationships (100) and (101) are only valid for antifoam free systems. 6.5.1.4 Influence of Cultivation Time and Operational Mode The two phase system properties depend considerably on the cultivation time and operation conditions. Immediately after inoculation, gas hold-up E G , diminishes, passes a minimum and then gradually and slightly increases. The behaviour of d and d s , resp., corresponds only partly to a mirror image of the E G course: 3 and d s increase at first, pass a maximum, a minimum and gradually and slightly increase (Fig. 17). The specific surface area, a, and volumetric mass transfer coefficient, kLa", have similar courses at the beginning of the cultivation: both of them diminish after the inoculation, pass a minimum and a maximum. After this maximum, a diminishes only slightly in contrast to kLa", which is considerably reduced as cultivation continues. Similar behaviour was also found with Candida boidinii on methanol, ethanol and glucose substrate 13). This course seems to be characteristic for antifoam free systems. The diminutions of E G , a and k L a a after inoculation are due to the increase

Characterization and Performance of Single- and Multistage Tower Reactors

141

Fig. 17 a—d. Variation of system properties during the growth of H. polymorpha on ethanol substrate 273 ' 274 ». Extended culture, c E10H = 5 g 1 _ 1 , wSG and EG were varied: a mean and Sauter bubble diameter, 3 and d s ; b specific interfacial area, a; c volumetric mass transfer coefficient, k L a; d cell mass, X, and oxygen transfer rate, Qo 2 , as a function of the cultivation time

142

K. Schugerl

Fig. 18a—d. Variation of system properties during the growth of H. polymorpha on ethanol substrate 2 7 3 •274"). After batch operation substrate shift was first carried out then extended culture operation with oxygen transfer limitation. w SG and E G were varied; a substrate concentration, C ElOH ; b mean and Sauter bubble diameter, d and d s ; c volumetric mass transfer coefficient, k L a; d oxygen transfer rate, Q 0 as a function of cultivation time, t

Characterization and Performance of Single- and Multistage Tower Reactors

143

of bubble size caused by considerable coalescence promotion. The origin of this phenomenon is not yet known. This coalescence promotion effect disappears after about 5 h, hence one can assume that it is caused by the yeast cells. They probably secrete surface active components to control their environment: At first to reduce the oxygen tension but after their adaptation to the new system, they consume them to increase the oxygen tension and OTR. The divergence between the courses of a and k L a", after they passed the maximum, can be caused by different phenomena: — by the diminution of k L caused by the coverage of the bubble surface by natural surface active substances, which are present in the cultivation media, — enrichment of small bubbles in the two-phase system. The gradual increase of E G with cultivation time seems to indicate that an enrichment of small bubbles exists. Because of the long small bubble retention time in the column, their oxygen content is totally exhausted. Thus, they no longer contribute to k L a. Hence k L a is reduced with increasing small bubble hold-up. The coverage of the interface by surface active substances (fatty acids, denaturated proteins) can only play a significant role for small bubbles with long retention times, since their enrichment at the interface needs considerable time. Since an interface covered by denaturated protein has high surface viscosity and viscoelasticity 2 1 3 ' 2 1 4 )

• v . •s

\

\

i -

10

15

20

25

h

30

\\ > 1

V

Fig. 19. Substrate concentration, S, and volumetric mass transfer coefficient, k L a, as a function of the cultivation time, t, employing substrate ethanol 1 6 ' 2 8 2 ) • S; o k L a

\\ N-o-Si

10

15

20

25

h

30

144

K. Schugerl

the coalescence of these bubbles is strongly inhibited. Therefore, one can presume that the formation of small bubbles and their surface coverage mutually intensify each other. Thus, the divergence between a and k L a, after they surpassed their maximum, is caused by small bubble formation and enrichment which is accelerated by their surface coverage, but not by the surface coverage of the medium and large bubbles. Both kLa" and a are strongly influenced by the concentration of substrates, which have coalescence suppressing characteristics. This substrate effect is adequately demonstrated in several publications (see 6.5.1.2). From Fig. 18 this effect can clearly be recognized. In the first phase of the cultivation, k L a" and a increase, as has already been discussed. At constant and high ethanol concentration a remains high ( > 1200 m - 1 ) as well as kLa" (>1000 h" 1 ) up to t = 25 h (Fig. 17). A reduction of the ethanol concentration, C EtOH , for a short period of time, causes a rapid diminution of a and kLa" to fairly low values (a ^ 600 h - 1 , kLa* ~ 400 h - 1 ) . If the substrate limitation prevails for longer time, a and k^ 01 are reduced even further, e.g. to kLa" ~ 200 h " 1 (Fig. 19). 6.5.2 Dependence of Turbulent Properties on Space and Operating Parameters In tower reactors the local mean liquid velocity profiles, U, are considerably nonuniform. U reaches high positive values in the column center and high 110 cm s 1 100

/

80

//

/r

60

/

¿0

D 20

\

\

1,

ii ?

v

-20

-40 -50

A - 6 - 4 - 2

0

2

4

6 cm

Fig. 20. True mean liquid velocity profiles, U, in tower reactors using porous plate 5 nm in pore diameter, and 1 % methanol solution, X = = 149 cm, wSL = 1,8 cm s" 1 at different superficial gas velocities, wS G 261> W x wSG = 2.67 cm s" 1 1 W ° S G = 5.35 cm s" v wSG = 8.02 c m s " 1

Characterization and Performance of Single- and Multistage Tower Reactors

145

negative values in the wall range. These profiles are influenced by the aerator type, distance from the aeration rate, wSG, as well as by the medium composition. At large distances from the aerator, the local mean velocity profile does not depend on x. When using porous plate gas distributors, for which d p d e , the local mean velocity profiles are flat in the homogeneous flow range ( < 6 c m s _ 1 ) and become considerably nonuniform in the heterogeneous flow range ( > 6 c m s - 1 ) (Fig. 20). With increasing hole diameter of perforated plate aerator (increasing d p ) these differences gradually diminish. When using a perforated plate 1 mm in hole diameter the differences between liquid velocity profiles in homogeneous and heterogeneous flow ranges are less, but still significant (Fig. 21) and with a perforated plate 3 mm in hole diameter, these differences completely disappear (Fig. 22). In the last system d p Si d e , thus no coalescence is possible and the size of the bubbles is nearly uniform. Therefore, as long as the bubbles can carry the gas which was fed into the system, homogeneous flow prevails. Only at high gas flow rates at which the bubble velocities are not high enough to carry the gas amount, bubble aggregates are formed which rise

K /*/ i f\

90 cms"'

A

80

60

40

n

J \ h *

20

1




/

\

\i

M \ v\

X

X

\

h -20

\

Fig. 21. True mean liquid velocity profiles, U, in tower reactors using perforated plate 1.0 mm in hole diameter and distilled water, x = 21 cm, wSL = 1.8 cm s" 1 at different superficial gas velocities 2 6 U (For symbols see Fig. 20)

-40 -50

-6

-4

-2

r

0

•

2

4

6 cm

146

K. Schiigerl

80

60 / X

*w

¿0 \\c

20

-20

§ - 2

6 cm 8

Fig. 22. True mean liquid velocity profiles, U, in tower reactors using perforated plate 3.0 mm in hole diameter and distilled water, x = 149 cm, w SL = 1.8 cm s " 1 at different superficial gas velocities 261) . (For symbols see Fig. 20)

with much higher velocities than individual bubbles and the transition to a heterogeneous state occurs. When employing porous plates, d p d e . Bubbles grow by coalescence, hence bubbles with broad diameter distribution are formed. Large bubbles are enriched in the column center and have much higher rise velocities than small ones. Since a close relationship exists between bubble velocity and mean liquid velocity 258) , high local liquid velocities are attained in the column center. In systems with a perforated plate 1 mm in diameter, the difference between d p and d e is fairly small. Thus the coalescence effect is moderate; bubble size distribution is fairly narrow. Therefore the mean local liquid velocity profile is less dependent of gas velocity, than the one with a porous plate. Furthermore the velocity maximum in the column center is smaller than the one with a porous plate, but larger than the one with a perforated plate 3 mm in hole diameter. The profile of the local mean velocity, U, is not influenced appreciably by the absence or presence of the bacterial cells, or by batch or continuous operation (Fig. 23). Also the radial profiles of turbulence intensity, u', and relative turbulence intensities, I, have the same shapes in the absence or presence of growing cells, in batch as well as in continuous operation (Fig. 24 and 25). The same holds true for the radial profiles of the relative mean gas hold-up, e G (Fig. 26), bubble diameters d and d s (Fig. 27) and bubble rise velocities, u B (Fig. 28) 261) . The radial profiles of macro time scales, t M , in these systems have parabolic shape with their minimum in the column center (Fig. 29). In cell free systems, this minimum is the broadest and has a much lower value (0.074 s) than those with growing cells (0.09 s in batch and 0.098 s in continuous operation) 1 0 2 ) .

Characterization and Performance of Single- and Multistage Tower Reactors

147

The radial profiles of micro time scales, xE, in these systems have similar shapes (Fig. 30). In the center of the column they are nearly constant. At the wall they steeply increase and at a distance of about 5 cm from the column center they pass a minimum. The dissipation time scale plateau in the column center is lower in the absence of cells (0.382 x 10" 3 s) than in their presence (0.4x 10" 3 s in batch and 0.42 x 10" 3 s in continuous operation).

Fig. 23. True mean liquid velocity, U, profiles in a single stage concurrent (air lift) tower loop reactor (Reactor A) using porous plate 17.5 |im in pore diameter and E. coli cultivation medium (Table 8). w SG = 2.17 cm s " 1 , T = 28 °C, x = 170 cm. Comparison of properties of the cell free system with those of cultivations 2 6 u o medium without cells a batch cultivation of E. coli • continuous cultivation of E. coli

Fig. 24. Turbulence intensity, u', profiles in a single-stage concurrent (air lift) tower loop reactor (Reactor A) using porous plate 17.5 ura in pore diameter and E. coli cultivation medium (Table 8). w SG = 2.17 cm s ' 1 , T = 28 °C, x = 170 cm. Comparison of properties of the cell free system with those of cultivation 261) . (For symbols see Fig. 23)

148

K. Schiigerl Fig. 25. Relative turbulence intensity, I, profiles in a single-stage concurrent (air lift) tower loop reactor (Reactor A) using porous plate 17.5 nm in pore diameter and E. coli cultivation medium (Table 8). wSG = 2.17 cm s - 1 , T = 28 °C, x = 170 cm. Comparison of properties of the cell free system with those of cultivations 2i(k) is shifted to higher energy dissipation rates but its shape and maximum position is not influenced, i.e. its efficiency remains the same (see curves 1 and 4 in Fig. 34). The one-dimensional power spectrum, E^n), and the energy dissipation spectrum, i>!(n), change only slightly in a radial direction in the tower 2 6 1 \ To describe these variations, different characteristics of these spectra were used. Sometimes the power spectrum can be described by a particular function in a wide range of n, e.g. Ej(n) ~ n~ 2 . In this case, the amplitude of the measured spectrum can be

152

K. Schiigerl

10*

10°

E,(k) 70"'

70"'

10

10-'

70"'

10° k/2n

cm-'

10 2

Fig. 33. One-dimensional power spectrum, E^k), in a tower reactor using porous plate 5 |im in pore diameter and 1 % methanol solution 2 6 0 , 2 7 3 ) . (For symbols see Fig. 32)

characterized by its distance from a reference straight line (in a log-log plot) and the frequency range of the validity of this particular function. Such a description of Ei(n) is sometimes possible in reactor B by employing E. coli medium (Table 8). In this case, E^n) ~ n~ 2 for a wide range of n. Thus the radial variations of the frequency range, n E , and amplitude, A E , were considered. In the absence of cells in the tower center, n E seems to have a maximum and A E a minimum. In the presence of cells (batch cultivation of E. coli), n E and A E vary only slightly in the tower cross section. Unfortunately, E,(n) ~ n " b generally holds true, where b is a function of n. With increasing n, b also increases. Thus this simple characterization cannot be employed. The energy dissipation spectrum must have a maximum. If the measuring technique allows one to find this maximum, Oi(n) can be characterized by the frequency at which this maximum occurs, n®, and by the height of the maximum, A^. Again in reactor B, n® and A® could be determined by employing E. coli medium (Table 8). In the presence of cells (in batch as well as in continuous cultures), n 0 and A^ are uniform in, the tower cross sections. In the absence of cells, A,,, is uniform and ni(k) in a tower reactor at wSL = 1.8 cm s _ 1 and r = 0 (at the center) using 1 % methanol, different aerators and superficial gas velocities 260,273) . | 1 porous plate 5 jim 2 perforated plate 1 mm > wSG = 2.67 c m s " 1 3 perforated plate 3 mm J wSG = 5.35 c m s " 1 4 porous plate 5 |im

In Reactor A and at further distance (x = 170 cm) from the aerator, n E and n® seem to be fairly uniform; A© seems to have a maximum in the center, A E seems to oscillate. At an intermediate distance (x = 30 cm) from the aerator, the position dependence of these parameters is not clear. In Figs. 35 and 36, one-dimensional power spectra and energy dissipation spectra are shown which were measured at x = 30 cm and in the center (r = 0) of Reactor A using a porous plate 17.5 |im in pore diameter and E. coli cultivation k medium (Table 8). 2 In Fig. 35, the dotted line represents Ej(n) ~ n~ . One can recognize from Fig. 35 that between model medium and batch cultivation only a slight difference exists. Both of them have a course of about E^n) ~ n - 5 / 3 in the range n = 8 to 150 Hz and for n > 150 Hz they follow the course E x (n) ~ n~ 2 . Fig. 36 again shows the slight differences between biological and model media for n = 20 to 500 Hz. One can recognize from this figure that the n-value at which ®i(n)'s maximum prevails, n®, is higher than 500 Hz, the upper limit which can be attained by the employed technique. In Fig. 37 one-dimensional power spectra are plotted which were measured in Reactor B (distance from Tray: 8 cm, r = 0 cm) employing E. coli medium (Table 8). The dotted line again represents the function Ej(n) ~ n - 2 . There are significant differences between the cellfree systems, batch and continuous cultivations of E. coli. In the relationship E^n) ~ n~ b , b continuously increases with increasing n from b ~ 0 at n ^ 7 Hz to about b ~ 3 at n = 500 Hz. The large differences between bio-

154

K. Schiigerl

50

150

Hz

500

Fig. 35. One-dimensional power spectrum E^n) in Reactor A using porous plate 17.5 |im in pore diameter and E. coli cultivation medium (Table 8). wSG = 2.17 cm s - 1 , T = 28 °C, x = 30 cm, r = 0 cm (tower center). Comparison of properties of the cell free system with those of batch culture 261 »; : E(n) ~ n~ 2 . (For symbols see Fig. 29)

logical and model media can be recognized from Fig. 37 b. In the absence of bacteria, n 0 > 500 Hz and thus it could not be determined. In a batch culture, n® and A® are higher than in a continuous culture. The drastic changes of ®i(n) in the presence of cells are probably due to the interaction of the foam with the stage separating trays. The low efficiency of Reactor B, when employing biological media, can be recognized from the low values of n® and A® in Fig. 37 b (see also Chapter 7.2.2).

7 Reactor Performance 7.1 General Considerations Bioreactors can be operated according to different strategies depending on the structure of the production costs. (Finishing costs are not considered here).

Characterization and Performance of Single- and Multistage Tower Reactors

•155

The following cases are considered : A. Substrate conversion, U s , should be maximized, if the raw material (substrate) costs dominate B. Productivity, Pr, should be maximized if investment and running costs (with exception of substrate) are high C. Oxygen conversion, U 0 , should be maximized, if oxygen transfer costs are important D. Specific power input should be minimized, if energy costs (for gas dispersion and cooling) are high. 7.1.1 Case A (Maximizing Substrate Conversion) The lower the intensity of longitudinal medium dispersion, the higher the conversion in continuous bioreactors. However, at low dispersion intensity cell washout occurs. 140 db

3

* *

130

x/V

f l x 9; ff xoo

120 xy

ff

110

c ^100

7

J 90

80

yJ

(

/ X /

0

3

10

50

150

Hz

500

Fig. 36. Energy dissipation spectrum, Bo0pt, i.e. Yopt > 0. On the other hand for Bo ^ Boopt, yopt = 0Fig. 41 shows how the Bo-number influences yopt. Near Boopt, yopt considerably depends on the Bo-number. At large Bo-numbers, yopt approaches a constant value and C s varies with Bo only slightly. Because ofjthe non-linearity of the Monod model, Yopt also exists slightly below Boopt. For comparison, the C s values for CSTR and for y = 0 with BOopt are also plotted in this figure. The transition from CSTR to dispersion model with Boopt and y = 0 improves (reduces) C s by 64 %. The introduction of the medium recirculation at, e.g., Bo = 15, further improves (reduces) C s by 33 %. Altogether a 76 % improvement of C s can be achieved with regard to CSTR. From Fig. 38 one can recognize that for this case (Bo = 1 5 ) and for y = 0 washout would occur. Washout can be avoided by recycling and C s minimized and/or U s maximized. In Fig. 42, an example is shown for Cs as a function of y at Bo = 7 and K = 1 for different Da-numbers. Since Bo > Boopt, y opt 's exist. For Da ^ Da crit = 3.41, yopt = oo (CSTR). For Da > D crit , y opt 's exist, at which C s is considerably improved. Fig. 43 shows yopt as

0.7 J

«—I—«—I I I » » I

1—I—I—I I 1111

0.6-

0

0.1

1—I—I—I 1111

Bo = 0.5 Da= 2 K = 0.5

• • • ' ... I I 0.5

1 1 1 I I IHI

1

5

Y

10

1 1 11 I I 50

100

Fig. 40. Dimensionless substrate concentration, C s , as a function of the medium recycling ratio, y. Comparison of different reac17 295) Itor I types '

160

K. Schiigerl

a function of the Da-number for two different K-values. For Da > Da crit , yopt diminishes with increasing Da. In Fig. 44, yopt is plotted as a function of K for three different Da-numbers. For K > K c r i , yopt diminishes with increasing K.

Fig. 41. Optimum recycling ratio, Y o p „ as a function of the Bo-number. Comparison of different reactor types' 7 - 2 9 5 1 Bo

7.1.2 Case B (Maximizing cell mass productivity) It is obvious that maximum cell productivity could be attained, if the cells could be cultivated in the nonlimited growth range, i.e. with |i m . It is well-known that the nonlimited growth in CSTR with sterile feed in a steady state is unstable 298) . The same is true of tower reactors with longitudinal dispersion 2 8 5 , 2 8 6 ) as well as with longitudinal dispersion and medium recycling 17,295) . However, it is possible to maintain unlimited growth at the entrance of the tower reactor and substrate limited growth at its exit, and by that to achieve high productivity and high substrate conversion in a stable steady state operation 21) . In this case the reactor behaviour can be described by the substrate limited rate equation. There is a formal analogy between autocatalytic reactions and cell growth with substrate limitation. Bishoff has shown that the maximum growth rate can be attained, if one uses a combination of a CSTR and PFR. The size of CSTR has to be chosen so that growth rate has a maximum in it 299) . The size of a CSTR can be avaluated graphically by plotting the inverse growth rate R £ _ 1 as a function of the cell mass concentration, X. Figures 45 and 46 show such plots, where

Characterization and Performance of Single- and Multistage Tower Reactors

161

the dimensionless growth rate is

Rx =

Rx

(105)

—— , X 0 + ^X/S o

the growth rate according t o M o n o d *

Rx =

XS

(106)

K. + S

Fig. 42. Dimensionless substrate concentration, C s , as a function of the recycling ratio, y, at Bo = 7, K. = 1 and for different Da-numbers 1 7 , 2 9 5 ) 50 100

Fig. 43. Optimum recycling ratio, yop„ as a function of the Da-number at Bo = 7 for different saturation constants K 17,2951

K . Schügerl

162 and the dimensionless cell mass concentration C,=

X

(107)

X0 + Y x / s S 0

One can recognize from Figs. 45 and 46 that R " 1 passes a minimum at a particular C , which is called C , crit . The CSTR-PFR combination only exhibits an optimum, if C x > C x c r i t . For C x ^ C x c r i l the C S T R is the optimum reactor. C x c r i t is given by 1 7 , 2 9 5 ) : (108)

crit

(K + 1) - j / K ( K + 1) ^ 0.5 i.e. for C x rg 0.5, C S T R is always the optimum reactor.

Fig. 44. Optimum recycling ratio, y o p „ as a function of the saturation constant, K at Bo = 7 and for different Da-numbers 1 7 •

295)

0.2

0.4 C,

0.6

0.8

1.0

Fig. 45. Reciprocal related growth rate, R , a s a function of the dimensionless substrate concentration, C„, at K = 0.125 for different n m „ . Combination of C S T R and P F R 1 7 ' 2 9 5 )

Characterization and Performance of Single- and Multistage Tower Reactors

163

If one uses a tower reactor with negligible longitudinal dispersion (PFR), the intensity of longitudinal dispersion can be controlled by employing medium recycling. There is an optimum recycling ratio, Yopt, at which R x has its maximum 17,295> . If one fixes the dimensionless cell concentration at the exit of the reactor, CXj F , and looks for a reactor, which yields the smallest necessary volume, VR, and mean residence time VR of the medium: T = or Da = T|IM, one finds that there is a PFR-loop combination with yopt at which Da has a minimum (Fig. 47), if C xF > C xFi crit , where (109)

C l F , c r i t = (K + l ) - l / K ( K + l ) For C I F ^ C xF

0

0.001

0.2

y opl = oo(CSTR)

cril)

OA

C, •

0.005 0.01

0.6

0.8

0.05 0.1

Fig. 46. Reciprocal related growth rate, R " 1 , as a function of the dimensionless substrate concentration, C x , at |i max = 0.5 h " 1 ( ) and variable K, as well as at n m = 0.3 h " 1 and K = 0.05 ( )17'295>

1.0

0.5

1

y

5

10

50 100

500 1000

— -

Fig. 47. Da-number as a function of the medium recycling ratio, for P F R with loop at K = 0.125, CXo = 0, Cv F = 0.95 17 ' 295 >

K. Schiigerl

164 For C xF > Cx, F crit . Da = (1 + y) "iIn

1 + y

r| + y

•h K iIn ^ + Y r| + y.

(110)

where c =

CXF 1 -Cx c = 1-c, yopt considerably depends on the exit cell mass concentration, C xF (Fig. 48). With decreasing C xF , yopt increases, and for C xF g C xFcri t, yopt = oo. yopt also depends on K: with increasing K, yopt diminishes 17). When employing a tower reactor with longitudinal dispersion and medium recycling, the relationships of case A can be used, since the exit cell mass concentration, C xF , can also be written as C„F

—

Y X /s(S 0 - S.) + X 0 X0 + Y x / s S 0

X0 + Y x / s S e

S0 - Se

= U.

(Ill)

Thus, reactor optimization with regard to U s (Case A) is also optimization with regard to C xF .

CxF

Characterization and Performance of Single- and Multistage Tower Reactors

165

7.1.3 Case C (Maximizing Oxygen Conversion) T o maximize the oxygen conversion, U D ,

UD = °

D O L o + OTR - DO L e OUR - = , DO L o + OTR D O L o + OTR

(112)

where 0 L o is the dissolved oxygen concentration in the medium at the medium entrance 0 L e the dissolved oxygen concentration in the medium at the medium exit.

OTR = J

z=0

kLa(0*-0L)dz

z=1

OUR =

.

z=0

1 — nx dz, 1 Y x/o

the growth must be oxygen transfer limited, at least at the medium exit. For a CSTR, Eq. (112) reduces to _ DOL0 + k L a ( O * e - O L e ) - D O L e Ur» — DOLo + kLa (0*e - 0 L e )

nX/Yx/0 DOLO + kLa (O*. -

0Le)

(113) For a given 0 L o and OTR, the oxygen conversion approaches unity This leads to a similar problem which was considered in Case A. The use of dissolved oxygen would be in a P F R . However, because of the medium recycling is necessary. In general, a type A tower loop longitudinal mixing and medium recycling can be used.

if 0 L e —> 0. most efficient cell washout, reactor with

However, the theoretical treatment of this system is more complex than the one considered in Case A, since oxygen is transfered into the medium along the column. Particular systems (batch and fed batch tower loop reactors in oxygen transfer limited growth range) were treated in Refs. 1 8 1 • 269 ~271>. In Chapter 5.2, the batch growth is considered.

By means of medium recycling it is possible to decouple the longitudinal dispersion of cell mass and substrate concentrations from the dissolved oxygen concentration. If the dissolved oxygen concentration, O l , reduces to zero in the loop, then at high recirculation ratios, y, the CSTR prevails with regard to the cell mass and substrate concentration and the P F R with regard to dissolved oxygen concentration 2 1 ) . Again to too high y values, the dissolved oxygen is also recycled and its conversion is reduced. One can assume, that similar to Case A, Bo opt as well as y opt exist, at which 0 L e has a minimum, and U Q has a maximum. However, this problem has not been treated quantitatively yet.

166

K. Schügerl

7.1.4 Case D (Minimizing Specific Power Input) In the oxygen transfer limited growth range the cell productivity is controlled by the OTR. In a CSTR, for the maximum productivity, Pr m , Eq. (114) holds true: Pr m ~ Y x / 0 k L a(0£ - Of)

(114)

D(0 L o - OL) « k L a(Oi - OD

(115)

since usually

Because of the low oxygen solubility, the maximum driving force (Of — Of) is low. Y x / 0 is determined by the microorganism and the substrate. Thus Pr m can usually be controlled by k L a. Since the variation range of k L is narrow, the specific gas/liquid interfacial area, a, is the main controlling parameter. According to Eq. (35), a depends on E G and d s . Since E G is also a function of d8, the Sauter diameter is the primary variable. d s depends in tower reactors on the primary bubble diameter, d p , and on the coalescence rate, R c , for which a simple linear relationship (106) is assumed: Rc = ^

- k c (d e - d s ) E g

(116)

In Eq. (116) E G kc

is a function of d s , is the coalescence rate constant, which is strongly influenced by the medium character, de the dynamic equilibrium bubble diameter, which is constant in a single stage tower, except the aerator range, and (dc — d s ) the driving force for the coalescence, which approaches zero with increasing time. The secondary conditions are given by ds = d p

for t = 0 , (at the aerator, x = 0)

(116a)

ds = de

for t -» oo (in large distance from the aerator x -> oo)

(116b)

The primary bubble diameter, d p , is controlled by the primary gas dispersion at the aerator. Since the gas dispersion is caused by turbulence forces also in tower loop reactors 280) , a close connection between the local turbulence energy dissipation rate and the local gas dispersion rate is expected. However, only relationships between the integral values of the energy dissipation rate, the integral specific power input, P L /V L , and the mean value of dynamic equilibrium bubble diameter, d e , are knowrT(e.g. 301, for a physical explanation see Chapter 4.3)

where Q is a constant. One can also employ Eq. (117) for the primary gas dispersion

Characterization and Performance of Single- and Multistage Tower Reactors

167

whire E is the loeal energy dissipation rate in the volume of primary gas dispersion, Vd the volume of the primary gas dispersion and C 2 a constant. Equation (118) does not consider the differences in efficiencies of the aerator. It assumes that the fraction of microeddies is the same in the turbulence flow and it does not depend on the aerator. However, there is strong evidence that aerators have very different efficiencies 302) and these differences are due to the different energy dissipation spectra of turbulence, which are produced by these aerators 1). To minimize the specific power input, the aerator efficiency has to be maximized. This holds true for primary dispersion. To minimize the specific power input, P L /V L , the coalescence rate, R c , has to be minimized as well. Different strategies for minimizing R c were considered in Ref. 303) .

7.2 Particular Tower Loop Reactors In Chapter 7.2, particular reactors will be considered separately and in Chapter 7.3, compared. 7.2.1 Single-stage Concurrent Tower Loop Reactor (Reactor A) T w o microorganisms were cultivated: Hansenula polymorpha

and Escherichia

coli\

the yeasts in batch, fed batch and extended culture in the absence and presence of antifoam agents (Desmophen, soy oil), and the bacteria in batch and continuous culture in the presence of antifoam agent (Desmophen). The H. polymorpha cultivations were computer simulated by means of a dispersion model (see 5.2) considering the space dependence of k L a according to Eq. (97) and calculating the longitudinal liquid dispersion coefficients, D L , and gas dispersion coefficients, D G , by means of Eq (119), which was recommended by Badura et al. 293) >0.33 ,D F (t) = D?(w^(t))°-

(119)

and by Eq. (120), which was recommended by Mangarz et al. 294) • : D G (t) = Do(w|(t)),3.0

In Eqs. (119) and (120): Di! = 2.4 x 10 - 4 m 2 s" 1 (dR x 102 m" 1 ) 1.4 and Dg = 5.0 x 10~8 m2 s _ 1 (dR x 102 m" 1 ),1.5

(120)

168

K. Schügerl

where wf G (t) is the dimensionless superficial gas velocity at the gas entrance with regard to wSG = 0.01 m s - 1 , w| G (t) wG(t) = the dimensionless effective gas velocity at the gas entrance, and Eg dR the tower diameter = 0.15 m In Fig. 49, some measured and calculated dissolved oxygen longitudinal concentration profiles are shown. These profiles were evaluated by the dispersion model (Chapter 5.2) and by Eqs. (97), (119) and (120) for extended culture of H. polymorpha at a constant ethanol substrate concentration (S = 5 g l _ 1 ) . By fitting the calculated profiles to the measured ones k L a E , kLa" and K ST in Eq. (97) were identified. The specific surface area, a, was calculated by Eq. (35), by means of measured E G and d s values (see Chapters 4.3.1 to 4.3.4). The mass transfer coefficient, k L , was evaluated by Eq. (117):

According to Eq. (121) it was assumed that the geometrical surface area, a, calculated by Eq. (35) and the mass transfer active area a, in k L a are identical. Non-limited and oxygen transfer limited growth The measurements employing H. polymorpha 16,153,15*,155,181,282) indicated that at the employed medium recycling rates (VR = 1000 to 2000 l h _ 1 ) the tower

1.00 Fig. 49. Longitudinal concentration profiles of dissolved oxygen, D O C (mg 1 ') in Reactor A during H. polymorpha cultivation at different cultivation times, t. Substrate concentration: S = 5 g 1 _ 1 154) ethanol . 1 m t = 2.0 h 5 x t = 13.0 h 2 A t = 6.0 h 6 • t = 14.0 h 3 v t = 10.0 h 7 a t = 17.0 h 4 + t = 12.0 h 8 o t = 25.0 h

Characterization and Performance of Single- and Multistage Tower Reactors

169

reactor exhibited CSTR behaviour with regard to the cell mass, X, and substrate, S, concentrations. The longitudinal concentration profiles of dissolved oxygen are nonuniform and can be described by a dispersion model with particular Bo-numbers and space dependent k L a. The profiles were fairly uniform in nonlimited growth due to the low oxygen uptake rates (curves 1 to 5 in Fig. 49), and in the strongly oxygen transfer limited range due to the high oxygen uptake rate (curve 8). At the beginning of oxygen transfer limitation (curve 7 in Fig. 49 and curve 1 to 3 in Fig. 50), the nonuniformity of the profiles is most significant. The dissolved oxygen was not recycled in the oxygen transfer limited growth range, because it was consumed in the loop as can be seen in Fig. 51 (upper part), in which the dissolved oxygen saturation is plotted as a function of the cultivation time, t, measured for Hansenula polymorpha on ethanol substrate at three different positions. At the end of the loop (z = —0), no oxygen could be detected even at t = 14 h, that is before the oxygen transfer limitation begins (t = 15 h, Fig. 51 lower part). Also when using E. coli on glucose substrate 1 7 ' 3 0 5 ) , the 0 2 -profiles were flat in nonlimited growth (curves 1 and 2 in Fig. 52). The dissolved oxygen concentration at the end of the loop (z = 0) already approached zero at t = 3.5 h (curve 3 in Fig. 52).

Fig. 50. Longitudinal concentration profiles of dissolved oxygen, DOC (mg 1 *) in Reactor A during H. polymorpha cultivation at different cultivation times, t. Substrate concentration: 5 g 1~' ethanol I 6 ' 1 5 4 ) . . | under strong oxygen 1 o t = 17.0 h 2 a t = 19.0 h } transfer limited 3 m t = 28.0 h J growth

170

K. Schiigerl

Fig. 51. H. polymorpha cultivation in Reactor A employing substrate ethanol in extended culture operation. Substrate concentration: 5 g P 1 kept constant by substrate feed. Aeration rate 0.55 vvm. Upper part: relative saturation of dissolved oxygen as a function of the cultivation time. Longitudinal position of the 0 2 -probes: • z = —0 (just below the aerator) • z = 0.09 (at the aerator) A z = 0.90 (at the tower head) lower part: variation of the cultivation with time • substrate uptake rate (g 1 _ 1 h - 1 ) o (dry) cell mass concentration, X (g l" 1 ) A oxygen uptake rate (g 1 _ 1 h - 1 ) v C 0 2 production rate (g 1"' h " 1 )

By using the dispersion model as well as Eqs. (97) and (119), the dissolved oxygen saturation profiles were calculated and fitted to the measured ones, thus k L a E , kLa" and K ST were identified. In Fig. 53, kLa® is shown as a function of the cultivation time for H. polymorpha and ethanol substrate 1 6 , 1 5 3 , 1 5 4 ' 1 8 1 ) . After inoculation, k^® drops to low values, then quickly increases, at t ~ 8 h, passes a maximum and diminishes at first fairly rapidly and after 14 h, gradually. Because the

Characterization and Performance of Single- and Multistage Tower Reactors

171

variations of the local d s and EG-values were slight in the tower (except for the aerator range, in which k L a E drops to k L a a ), it was possible to characterize the two phase flow in the tower with local measurements of d s and E G . In Fig. 54, the specific interfacial area, a, calculated by Eq. (35), is plotted as a function of t for same run for which k L a a is shown in Fig. 53. An initial drop, then a quick increase is common for kLa* and a. However, k L a a passes a maximum earlier and drops rapidly, while a attains its maximum later and diminishes only slightly. The mass transfer coefficient, k L , was calculated by Eq. (121) (Fig. 55). For t si 4 h it is constant, but soon after it gradually diminishes for t ^ 15 h, somewhat quicker, then-slighter 1 6 , 1 5 3 , 1 5 4 , 1 8 1 ) . It is interesting to compare these properties of

Fig. 52. Longitudinal concentration profiles of dissolved oxygen with regard to the saturation (%) in Reactor A during E. coli cultivation in batch operation. V„ = 1100 1 h " 1 , at different cultivation times 17 ' 315 >. o t = 30 min a t = 115 min • t = 210 min

Z 1800 h"1 1600 1400

/

/

/

1200

o 1000

o o

X

800 600

0

5

10

15

t

20 —

25

h

30

Fig. 53. Volumetric mass transfer coefficient, k L a, as a function of the cultivation time in Reactor A during the cultivation of H. polymorpha on ethanol substrate, S = 5 g l - 1 . Aeration rate: 0.55 w m 1 6 ' 1 5 4 »

172

K. Schügerl

1600 1400

o

^

o

o /

1200

/

1000

\

°

1

/

Fig. 54. Specific interfacial area, a, as a function of the cultivation time, t, in Reactor A during the cultivation of H.polymorpha on ethanol substrate, S = 5 g 1 _ 1 . Aeration rate: 0.55 w m 1 6 - 1 5 4 »

1

800 600

o 400

10

15

20

25

h

30

t

0.06 cms" 1 0.05 0.04 0.03

"

A

\\

\ „

I o r>-SL"a-!r-

Fig. 55. Mass transfer coefficient, k L , as a function of the cultivation time, t, in Reactor A during the cultivation of H. polymorpha on ethanol substrate, S = 5 g 1 _ 1 . Aeration rate: 0.55 vvm 1 6 ' 1 5 4 )

0.02 0.01

10

15

20

25

h

30

an antifoam free medium (//. polymorpha) with those of a medium (E. coli) in the presence of an antifoam agent (Desmophen). (The k L a values were calculated for the latter by means of 0 2 balance) 1 7 , 3 0 5 ) . In Fig. 56, k L a and k L are plotted as a function of t. Both diminish with increasing cultivation time. Their courses are very similar because of d s and E G , thus a varies only slightly with t. Also their spacial variations in the tower are slight (Fig. 57). In the antifoam free system k L a drops from ca. 1600 to 800 h " 1 (Fig. 53), and in presence of Desmophen from 2100 to 650 h " 1 (Fig. 56). In both of these systems a varies from ca. 1400 to 1200 m " 1 (Fig. 54 and Fig. 57). However, k L is much lower in the presence of Desmophen (k L SÌ 6 x 10~3 cm s - 1 for t > 6 0 m i n ; Fig. 56) than in its absence (k L a; 5 0 x l 0 " 3 c m s - 1 f o r t < 4 h a n d k L 2 0 x 10~3 cm s" 1 f o r t > 15h;Fig.55). At low aeration rate, at which coalescence does not yet dominate, one can determine the ratio of k L values from the corresponding ratio of k L a values. Such measurements were carried out after the cultivation was stopped in a small twin bubble

Characterization and Performance of Single- and Multistage Tower Reactors

t

173

(mini

Fig. 56. Mass transfer coefficient, k L , and volumetric mass transfer coefficient, k L a, as a function of the cultivation time in Reactor A during the cultivation of E. coli in batch operation. V R = 1051 1 h " 1 1 7 ' 3 0 5 )

1600 a K00 [m-i] 1200 -

1000 1.5

V

s

[mm]

• 1 1.0

d

&

O

B

0.1

0.2

0.3

0À

0.5

s

Q

t [min]

A

tk

325

Q

• •

a

505

»

685

0.8

0.9

O

0.5

d

&

0.6

0.7

1.0

Z Fig. 57. Specific interfacial area, a, mean bubble diameter, d B , and Sauter bubble diameter, d s , as a function of the longitudinal position in Reactor A during the cultivation of E. coli at different times in continuous culture at high recycling ratio 1 7 ' 3 0 5 )

174

K. Schiigerl

Fig. 58. H. polymorpha cultivation in Reactor A employing glucose substrate in extended culture operation. Substrate concentration, S = 9.2 g l ~ \ kept constant by substrate feed. Aeration rate: 0.55 w m 16'154>. upper part: dissolved oxygen concentration as a function of the cultivation time. Longitudinal position of the 0 2 -probes: (just below the aerator) • z = —0 • z = 0.09 (at the aerator) a z = 0.90 (at the column head) lower part: variation of the cultivation with time • glucose uptake rate (g 1 _ 1 h" 1 ) o (dry) cell mass concentration, X, (g l" 1 ) a oxygen uptake rate (g l" 1 h" 1 ) • C 0 2 production rate (g 1 _ 1 h - 1 )

Characterization and Performance of Single- and Multistage Tower Reactors

175

column 218) . These measurements yielded kL values for H. polymorpha which were by a factor of two higher than those for E. coli in the presence of Desmophen. In model media this antifoam effect is much larger (by a factor 5 to 6) 2 1 8 ) . Unexpected were the small d s values in the presence of Desmophen. The large amount of protein (meat extract) in the E. coli medium seems to partly compensate the antifoam effect. This can also be recognized by means of a definite foam formation tendency in spite of the presence of Desmophen. The courses of kLa" and a as a function of the cultivation time are rather different during H. polymorpha cultivations on ethanol and glucose substrate 16>153'154> 181>. When employing glucose, the dissolved oxygen saturation diminishes with increasing t as usual, but at 3.5 h passes a maximum and gradually increases (Fig. 58, upper part). This happens at the end of the exponential growth (Fig. 58 lower part). However the deviation from exponential growth was not caused by growth limitation, because sufficient substrate and dissolved oxygen were present (Figs. 58 and 59 upper part). The reduction of the growth rate is accompanied by diminutions of the yield coefficients Y x / 0 and YX/s and the cell mass productions (Fig. 59 middle and lower parts). Also ethanol is produced (Fig. 59 upper part). Obviously

12 •

/

" I

•

•

gi-

• Gluco.se (subs träte) o E thai oKproi 1uced)

lgl-1

«I

§


176

K. Schügerl

mgl

1.00 Fig. 60. Longitudinal profiles of dissolved oxygen (mg 1 ' ) in Reactor A during the cultivation of H.polymorpha on glucose substrate (S = 9.2 g I" 1 ) at different cultivation times, t 1 6 , l 5 4 ) ; 1 m t = 6.0 h X = 7.10 g l " 1 2 A t = 7.0 h X = 9.65 g l " 1 3 o t = 9.5 h X = 10.05 g r 1

800 h"' y 700

600

/

/

o

f

500

Of O /

400

300

200 X

Fig. 61. Volumetric mass transfer coefficient, k L a, as a function of the cultivation time, t, in reactor A during the cultivation of H. polymorpha employing glucose substrate, S = 9.2 g l - 1 . Aeration rate: 0.5 w m 1 6 ' 1 5 4 )

'

100 10

h 12

Characterization and Performance of Single- and Multistage Tower Reactors

177

some kind of repression has occurred due to glucose. (With the diminution of glucose concentration this effect can be reduced.) The longitudinal dissolved oxygen profiles are similar in this range to those measured in systems with ethanol substrate at the end of the exponential growth phase (Figs. 50 and 60). By fitting the calculated profiles to the measured ones the corresponding k L a a values were determined. Fig. 61 shows that kLa" increases with t. The same holds true for a (Fig. 62). This is caused by the increasing ethanol concentration. The k L value is nearly constant (Fig. 63). A comparison of kLa" values with ethanol and/or glucose substrate shows that the former k L a a is much higher (1400 800 IT 1 ) than the latter (150 700 h" 1 ) (Figs. 53 and 61). With increasing alcohol concentration in the glucose system the k ^ " values also approach 800 h _ 1 . The same is true for the specific interfacial area, a, with ethanol: a = 600 -» 1400 m " 1 (Fig. 54) and with glucose: a = 200 -> 1000 m " 1 (Fig. 62). k L values are also higher in the ethanol (Fig. 55) than in the glucose (Fig. 63) system but with increasing time and ethanol concentration they approach the same value (20 x 10~3 h - 1 ) .

Fig. 62. Specific interfacial area, a, as a function of the cultivation time, t, in Reactor A during the cultivation of H. polymorpha employing glucose substrate, S = 9.2 g 1~1. Aeration rate: 0.5 w m , 6 ' 1 M )

Fig. 63. Mass transfer coefficient, k L , as a function of the cultivation time, t, in Reactor A during the cultivation of H. polymorpha on glucose substrate, S = 9.2 g r 1 . Aeration rate: 0.5 w m 1 6 ' 10

h

12

178

K. Schügerl

The influence of ethanol concentration on kLa* is shown in Figs. 13 and 15. k^a" increases with increasing ethanol concentration. The influence of the superficial gas velocity on OTR is shown in Fig. 14 and on k L a" in Fig. 16. Both of them indicate that with increasing superficial gas velocity k ^ increases, passes a maximum, then diminishes. This is due to the increasing medium recycling rate, VR. In Fig. 64 k L a is shown as a function of VR during E. coli cultivation 17 • 305) . The reduction of k L a with increasing VR is significant. This is due to the reduction of E G . According to Reith 304) the mean relative gas holdup, (E G ) AL , can be calculated for liquid recycling by: (EG)AL =

Wo 2 w

-

r

+

s g

W

(122) s

where w s is the relative bubble swarm velocity with regard to the liquid velocity. Since in bubble columns (EG)B,C. is given by Eq. (123), (E g ) b .c. = —

ws

.

(123)

the reduction of E G due to air lift liquid recirculation is given by: (Eg)al (EG)b.c.

Ws 2 w

s g

+

• w

(124)

s

This reduction of E G is mainly caused by the increase of liquid velocity due to higher VR (see Eq. (127)). Substrate

limited

growth

In Fig. 65, the course of H. polymorpha cultivation on ethanol substrate is shown 16 - 153 - 282 > : ethanol feed rate, m EtOH , ethanol concentration, S EtOH , dissolved

Fig. 64. Volumetric mass transfer coefficient, k L a, as a function of the medium recycling rate, VR in Reactor A during the cultivation of E. coli in continuous operation 17, 305)

vR (I h-i)

Characterization and Performance of Single- and Multistage Tower Reactors

179

Fig. 65. Growth transitions during the cultivation of H. polymorpha in Reactor A employing ethanol substrate , 6 ' 2 8 2 ) . a ethanol feed rate, mE«jH; b ethanol concentration, S; c relative dissolved oxygen saturation, D O S ; O T R ; e (dry) cell mass concentration, X, and cell productivity, Pr, as a function of the cultivation time

oxygen saturation, 0 2 -sat., OTR, QQ2, cell mass concentration, X, and cell mass productivity, Pr, are plotted as functions of the cultivation time, t. The cultivation started with batch mode until the substrate was consumed. Then the cells were fed batch cultivated at first at a low feed rate. The feed rate then was increased step by step. Substrate limited growth prevailed. By increasing the ethanol feed rate the substrate limited growth turned into oxygen transfer limited growth at a given feed rate and the ethanol concentration began to increase as can be seen from Fig. 65. At t = 7.5 h the substrate was consumed. Accordingly the cell growth stopped, the productivity diminished to zero, the OTR also reduced and the dissolved oxygen saturation increased. Shorthly yfter t = 7.5 h the substrate feed was started at 0.71 g EtOH 1 _ 1 h" 1 . The cells began to grow again, Pr and OTR increased dissolved oxygen saturation diminished, but the substrate concentration remained at a very low level. No ethanol could be detected. By increasing the substrate feed rate, the growth rate, Pr and OTR increased and dissolved oxygen saturation

K. Schugerl

180

decreased. At every substrate feed rate a quasi-steady state value of these variables was attained. At t = 17 h the substrate feed rate was increased again, now from 1.37 to 1.63 g 1 _1 h - 1 . As a result the substrate limited growth turned into oxygen transfer limited growth. The dissolved oxygen saturation was reduced to its critical value, Ol/O*, below which oxygen could not be utilized by the organisms. Hence the maximum OTR was attained, which controlled the growth rate. Since under these conditions more substrate was fed than could be consumed, the ethanol concentration increased. The maximum productivity was attained and remained constant (1.05 g l" 1 h - 1 ) even with increasing substrate concentration. To make the transition from substrate limited growth to oxygen transport limited growth more evident, the productivity Pr, and OTR are plotted as functions of the dissolved oxygen saturation in Fig. 66. For dissolved oxygen saturation larger than 20 % substrate limited growth exists. Below this value oxygen transfer limitation prevails. One can also see how with increasing dissolved 02 driving force (1 — 0 L /0*) Pr and OTR increase. At the boundary between substrate limited and oxygen transfer limited growth a productivity of 0.95 g 1 _1 h" 1 can be attained. In the oxygen transfer limited range Pr = Prm = 1.05 g l - 1 h - 1 . In Fig. 67a, the longitudinal dissolved oxygen concentration profiles, which correspond to the run represented by Fig. 65, are shown for different cultivation times, t. Curves 1 to 4 are characteristic for nonlimited growth, 5 and 6 for substrate limited growth and 7 for oxygen transfer limited growth. The curves in substrate limited growth show that the dissolved oxygen concentration in the upper half of the column increased after its decrease in the lower half of the column. This peculiar behaviour can be explained by the change of the nonlimited growth in the lower column half to a substrate limited one in the upper column half. Hence the oxygen uptake rate diminishes in the upper column half. Since the dissolved oxygen concentration at the interface remained the same, the necessary

a 02 limited g rowth

1 i

\

2.0 is Irate lim ited growt h

f

1.2

o 0i 0.8

a

Qo? —' • Pr

¡0

OA

20 dissolved

30 ' / . of saturation SO 0? -

Fig. 66. H. polymorpha cell productivity and O T R in Reactor A as a function of the relative dissolved oxygen saturation, DOS, employing substrate ethanol 1 6 • 2 8 2 1

Characterization and Performance of Single- and Multistage Tower Reactors

181

Fig. 67 a. Longitudinal dissolved oxygen concentration profiles at different cultivation times in Reactor A during the cultivation of H. polymorpha on ethanol substrate. Profiles were calculated by assuming perfect mixing with regard to the substrate 16'282>. • / t = 1.5 h a 5 t = 8.0 h a 2 t = 4.0 h o 6 t = 9.0 h v 3 t = 6.0 h • 7 t = 12.0 h + 4 t = 7.0 h b. Longitudinal dissolved oxygen concentration profiles at different cultivation times in Reactor A during the cultivation of H. polymorpha on ethanol substrate. Profiles were calculated by taking the longitudinal concentration profile of substrate into account 1 8 1 ' 2 7 2 ) . 0 J t=1.5h v 6 t = 12.0 h substrate limited growth O 2 t = 6.0 h O 7 t = 14.0 h • 3 t = 7.0 h batch a 8 t = 17.0 h \ , .. . o 4 t = 7.7 h [ « 9 t = 20.0 h / °*ygen transfer limited growth • 5 t = 8.0 h

182

K. Schiigerl

oxygen could be supplied by a reduced driving force. As a result the dissolved oxygen saturation, 0 L / 0 * , increased. The symbols in Fig. 67 a represent the measured values and the curves the calculated profiles assuming CSTR behaviour in the tower with regard to the cells and substrate and a definite longitudinal dispersion with regard to the dissolved oxygen 1 8 1 , 2 6 9 ~ 2 7 1 ) . One can see that these assumptions do not hold for substrate limited systems. These profiles can only be described by employing a model with distributed parameters with regard to the dissolved oxygen and substrate concentration and lumped parameters with regard to the cells. Such models have been developed by Luttmann 181> and Scheiding 272) . The agreement between the courses of longitudinal dissolved oxygen concentration profiles measured and calculated by these models are also excellent (Fig. 67 b), if one assumes that along the tower a

Fig. 68. Growth transitions during the cultivation of H. polymorpha in Reactor A on substrate glucose 16>282>; a glucose feed rate, m GllI ; b glucose concentration, S; c oxygen transfer rate, QQ2; d relative dissolved oxygen saturation, DOS; e (dry) cell mass concentration, X, and cell productivity, Pr; as functions of the cultivation time, t 1 6 , 2 8 2 ) .

Characterization and Performance 6f Single- and Multistage Tower Reactors

183

change of nonlimited growth to substrate limited growth occurs. Analogous to the run plotted in Fig. 65, the glucose concentration was also varied in other runs 1 6 , 1 5 3 , 2821 (Fig. 68). After batch cultivation and substrate consumption the substrate feed was started at a low level and then increased step by step. Again by consumption of the substrate the growth rate, the productivity and OTR were reduced to zero and dissolved oxygen saturation increased. By increasing the substrate feed rate step by step quasi-stationary states were maintained. Between t = 2.5 and 12 h no glucose could be detected in the medium. Thus, substrate limited growth prevailed. Up to this point no difference between the behaviour of the ethanol and glucose systems could be observed. By increasing the glucose feed rate from 3.6 to 4.7 g 1~1 h _ 1 the substrate concentration began to increase. Analogous to the ethanol system one would expect, from this point on, oxygen transfer limited growth. However, dissolved oxygen saturation did not change, the OTR increased only slightly, but its maximum had not been attained. All these data indicate that the cell had enough oxygen ( 0 L / 0 * — 0.4) and enough substrate, i.e. non limited growth existed. However, the productivity did not increase, but even decreased probably due to glucose repression, which is accompanied by ethanol production. The maximum cell productivity (1.7 g l " 1 h" 1 ) was attained at a substrate feed rate of 3 . 6 g T 1 h " 1 under substrate limited conditions. The transition from substrate limited growth to oxygen transfer limited growth could not be achieved under these conditions. In Fig. 69, the longitudinal profiles of the dissolved oxygen are again plotted for different cultivation time, t, by employing glucose as substrate. Curves 1 to 4

z Fig. 69. Reactor / t 2 t 3 t 4 t

Longitudinal dissolved oxygen concentration during the cultivation of H. polymorpha A on substrate glucose at different cultivation times, t 1 6 , 2 8 2 ) . = 0.01 h 5 t = 7.5 h = 1.0 h 6 t = 8.5 h = 1.75 h 7 t = 14.5 h = 5.5 h

in

184

K. Schügerl

are characteristic for nonlimited growth, 5 and 6 for substrate limited growth and curve 7 for glucose repression. Analogous to the ethanol system in the nonlimited growth region, the dissolved oxygen concentration diminishes in the lower half of the tower and increases in the substrate limited growth region in the upper half of the tower. These longitudinal dissolved oxygen concentration profiles could also be described by a distributed parameter model with regard to the substrate and oxygen concentration. Figure 19 shows how kLa® varies during the run represented by Fig. 65. After its initial increase kLa® is reduced with diminishing ethanol concentration from 1600 to 220 h _ 1 . The latter prevailed in the substrate limited range. In the oxygen transfer limited range, aifd at detectable ethanol concentrations, kLa"* increases again and approaches a constant value (350 h - 1 ) . The course of a is similar to that of k L a". In Fig. 70, k L a" and a are shown as a function of the cultivation time for the run represented by Fig. 68. As one can see, no dramatic variations of k L a" and a can be observed. The gradual increase in k L a" and a for t > 4 h is due to increasing alcohol concentration. It is interesting to consider how the variation of the medium recirculation rate, VR, influences the longitudinal concentration profiles of the medium components in the tower. Typical cultivation conditions by employing E. coli in a continuously operated tower loop reactor at high medium recirculation rates, VR, are shown in Figs. 71 and 72 17 > 305 '. The steady state is quickly attained at high VR. A substrate

X , 4

o o

°

lo

D

c

o

Fig. 70. Substrate concentration, S, volumetric mass transfer coefficient, k L a, and specific interfacial area, a, during the cultivation of H. polymorpha in Reactor A on substrate glucose as function of the cultivation time, 1 16 - 282) ; • C o l u ; o k L a; A a

*

12

16

Characterization and Performance of Single- and Multistage Tower Reactors

185

Fig. 71. Cell mass concentration, X, and substrate concentration, S, during the cultivation of E. coli in Reactor A in batch and continuous operation as functions of the time, t. S 0 = 5.5 g l - 1 , D = 0.44 h" 1 , VR = 1400 Hi" 1 , y = 70 1 7 ' 3 O 5 )

Fig. 72. Oxygen and C 0 2 concentrations, x and xcc>2, in gas phase and respiratory quotient, R.Q., during the cultivation of E. coli in Reactor A in batch and continuous operation as functions of the o x^, • x ^ , O R.Q. time, t. (For operation conditions see Fig. 71) 1 7 , 3 0 5 ) ,

186

K. Schugerl

Fig. 73. Cell mass concentration, X, and substrate concentration, S, during the cultivation of E. coli in Reactor A in batch and continuous operation as functions of the time, t. S 0 = 5.4 g l " 1 , D = 0.37 h - 1 , VR = 0, y = 0 (no medium recycling) 17 3 0 5 ) ; • X at x = 25 cm a S at x = 25 cm a S in loop o X in loop

h1 0.2

X

=3.89/1

1 V B = 370 l/h i

•

K /„=

>

0.4

0.6

0.8

01/h

1.0

Fig. 74. Longitudinal profiles of cell concentration, X, and substrate concentration, S (in upper part) as well as dissolved oxygen saturation, DOS, during the cultivation of E. coli in Reactor A in steady state operation 17 3 0 5 ) . o X . a S , oDOS. (For operation conditions see Fig. 73); 1 • D O S at VR = 3701 h "

187

Characterization and Performance of Single- and Multistage Tower Reactors

conversion, U s , of ca. 100% is achieved at D = 0.44 h - 1 . On the other hand, in Figs. 73 and 74 cultivation conditions are shown in the absence of medium recirculation. During the batch phase the medium was recirculated: VR = 375 1 h " 1 (y = 22.2). After starting the medium feed: D = 0.37 h - 1 , in continuous operation, VR was reduced to zero. During the first 300 min of continuous operation the process variables are similar to those measured at high VR values. However, the cell mass concentration, X, which was measured at x = 25 cm (above the aerator), gradually increased (Fig. 73) due to the nonuniform longitudinal profile of cell mass concentration (Fig. 73). As one can see from this figure, which is characteristic for the end of this run, the cell mass concentration is 50 % higher in the lower half of the tower than in the upper half. In spite of this the substrate concentration remained uniform in the tower. The longitudinal concentration profiles of dissolved oxygen are considerably nonuniform. In the lower half of the tower high dissolved oxygen concentrations prevail (Fig. 74). They drop to very low values at about the half height of the tower and then gradually increase. When increasing the recirculation rate, VR, this profile becomes more uniform (Fig. 74). In the absence of medium recirculation, cell sedimentation occurs, which leads to oxygen transfer limitation in the upper half of the tower. This causes cell flocculation, which increases cell sedimentation. Thus sedimentation and flocculation are mutually amplified. Under these conditions the substrate conversion reduces to 88.5 % and the yield coefficient, Y x/S , to 0.8. 7.2.2 Ten Stage Concurrent Tower Loop Reactor (Reactor B) 17'315> Figures 75 and 76 show a typical run in the ten stage tower loop reactor by cultivating E. coli with D = 0.33 h - 1 and VR = 233 1 h " 1 . The courses of X and S are similar to those in the single stage tower by employing the same bacterium (Fig. 71), but in the ten-stage tower the steady state has not yet been achieved up to t = 800 min. The variation of X with t is not caused by the variation of the local cell mass concentrations (X's in the tower and in the loop are constant and equal), but

\

>-

9 1-1

'

....

ri continuous

m

. A

jmin

à

»—

600

t

Fig. 75. Cell mass concentration, X, and substrate concentration, S, during the cultivation of E. coli in Reactor B (ten-stage tower loop) in batch and continuous operations as functions of the time, t. S 0 = 5.6 g l " 1 , D = 0.33 h " 1 , VR = 233 1 h " 1 1 7 ' 3 0 5 ) ; • X at x = 25 cm A S at x = 25 cm o X in loop a S in loop

188

K. Schügerl

u o> I

o ii

III 1

IV V VI VII vm IX X i Fig. 76. Longitudinal profiles of cell mass concentration, X, substrate concentration, S, (upper part) during the cultivation of E. coli in Reactor B in batch and continuous operation 1 7 , 3 0 5 1 ; o X, S (continuous) upper part DOS at t = 40 min (batch) DOS at t = 220 min (conti.) lower part DOS at t = 270 min (conti.) DOS at t = 400—820 min (conti.). X are the stages counted from the bottom

by the variation of the overall cell mass concentration in the tower. The differences in S in tower and loop disappear after 600 min (Fig. 75). The uniform cell mass concentrations, X, and substrate concentrations, S, in the ten-stage tower are shown in Fig. 76. On the other hand, the longitudinal dissolved oxygen saturation profiles are considerably nonuniform. At high medium recirculation rate and batch operation the dissolved oxygen saturation diminishes with increasing stage number (counted from the bottom), but it remains higher than 0.5. Thus no oxygen transfer limitation occurs. In continuous cultivation the dissolved oxygen saturation drops to zero. With increasing cultivation time the numbers of stages increase in which the oxygen supply to the bacteria is not sufficient (Fig. 76). (The increase of the oxygen concentration in the 10th stage is caused by the mechanical foam destroyer at the top of the column, which also acts as an aerator). The inadequate oxygen transfer rate in the ten-stage tower is due to the formation of liquid free layers below the perforated trays, which acted as very ineffective gas distributors, because of their large (3 mm) hole diameters and free cross sectional area (6.52%). In addition, the efficiency of the stage separating trays probably deteriorated due to foam formation below the trays. The foam lamellae block the tray openings. Thus gas pressure increases. At a critical overpressure, the lamellae in one of the holes burst and the gas escapes through this hole of the tray to the next upper stage. Thus the tray actually acts as a single hole aerator plate which is especially ineffective. W h ' n reducing the medium recirculation rate the cells are enriched in the 1st stage. However, X is constant in the rest of the stages. In the tower no cell sedimentation occurs. The high X-values in the 1st stage are caused by cell

189

Characterization and Performance of Single- and Multistage Tower Reactors

sedimentation in the loop. The substrate concentration is nearly constant in the loop system. The longitudinal profile of the dissolved oxygen saturation is similar to that found at higher medium recirculation rate 17-305>. At a very low medium recirculation rate (VR = 16 1 h - 1 , y = 1) differences appear between X's and S's in tower and loop during batch cultivation (Fig. 77). At the beginning of the continuous culture, X is higher in the loop than in the 1st stage of the tower. With increasing cultivation time X in the loop gradually diminishes and in the 1st stage remains constant. After 550 min, X becomes higher in the 1st stage than in the loop (Fig. 77). During the crossing of these profiles and thereafter S increases considerably. However, S is always higher in the 1st stage than in the loop, because the feed is introduced into the 1st stage. It is not known why this crossing of X profiles occurs. No dramatic changes of XQ2-, XCQ2- and RQ-courses can be recognized during the X profile-crossing (Fig. 78). The longitudinal profiles of X, S and pH indicate (Fig. 79) that X has a slight minimum in the 2nd stage, and then slightly diminishes with increasing stage number. In the ten-stage tower loop the diminution of VR also reduces the substrate conversion, U s , and yield coefficient, Y x/S . 7.2.3 Single-stage Countercurrent Tower Loop Reactor (Reactor C)

87 88,89)

'

H. polymorpha was cultivated in reactor C in extended culture under fixed operation conditions : the liquid volume of the cultivation medium was 501, the aerated height 160 cm, the aeration rate 1 w m (wSG = 3.02 cm s _ 1 ), the liquid recirculation rate, VR = 200 1 h _ 1 and ethanol concentration S = 1 % and/or 0.8% per volume. At first antifoam free systems are considered. In Fig. 80, dissolved oxygen saturations in the middle of the tower and in the loop are shown as functions of cultivation time. The dissolved oxygen saturation in the loop is always much lower than in the tower. After 8 h no dissolved oxygen could be detected in the loop. Around this time the oxygen transfer limitation became effective and the specific growth rate

Fig. 77. Cell mass concentration, X, and substrate concentration, S, during the cultivation of E. coli in Reactor B in batch and continuous operations as function of the time, t. S 0 = 5.8 g l - 1 , D = 0.36 h" 1 , VR = 16 l h " 1 , y = 0 . 9 5 17'305>. • X at x = 25 cm A S at x = 25 cm o X in loop & S in loop

190

K. Schiigerl

5r 21 %

RQ

*co2 * 2

lOOfcontinuous

t

400

-

Fig. 78. Oxygen and C 0 2 concentrations, Xq2 and Xco2 in gas phase and respiratory quotient, R.Q., during the cultivation of E. coli in Reactor B in batch and continuous operation as function of the time, 1 17 ' 305 >. (For operation conditions see Fig. 77); a 0 2 , A C 0 2 , • R.Q.

7 6

91"

pH

d5 X,S O Û

III

0.2

0.4

IX

0.6

0.0

1.0

Fig. 79. Longitudinal profiles of cell mass concentrations, X, substrate concentration, S, and p H . value during the cultivation of E. coli in Reactor B in continuous operation 17'305>. (For operating conditions see Fig. 77) ; o X, A S, • pH; I—X are the stages counted from the bottom

began to decrease. Also the curves of X, substrate uptake rate and OTR began to deviate from the exponential course and evened off (Fig. 81). This transition was completed after 12 h and the growth became linear in the entire reactor due to oxygen transfer limitation of growth. A linear .plot of X as a function of t indicates linear growth of the cells during this phase. At this time maximum cell productivity of 0.71 g l - 1 h _ 1 and a maximum

Characterization and Performance of Single- and Multistage Tower Reactors

o io o S? 60

191

Fig. 80. Dissolved oxygen concentration as a function of the cultivation time, t, during the cultivation of H. polymorpha in Reactor C in absence of antifoam agent and employing substrate ethanol. (For operation conditions see Fig. 81) 87 ' 88 »; o loop; • in the middle of the tower

40

•o 01 "5 in 20 ,w O O) «f

Fig. 81. Extended cultivation of H. polymorpha in Reactor C in absence of antifoam agent employing substrate ethanol S = 1%, aeration rate: 1 vvm, medium recycling rate: VR = 2001b" 1 8 7 , 8 8 ) ; A cell mass concentration (g 1"') • OTR (g 1 _ 1 h" 1 ) o consumed ethanol (g l" 1 )

192

K. Schiigerl

OTR of 1.28 g h _ 1 were attained and the dissolved oxygen saturation in the middle of the tower amounted to 0.25. Under the given operation conditions the maximum productivity, controlled by the OTR, had been attained at this point. In the presence of antifoam agent Desmophen, |am was considerably reduced from 0.25 to 0.21 h - 1 , k L a from 250 to 160 h " 1 and Y x / 0 increased from 0.55 to 0.69. The results obtained in the presence of antifoam agents (Desmophen and/or soy oil) are discussed in 7.2.4. 7.2.4 Three-stage Countercurrent Tower Loop Reactor (Reactor D)

87 88 89)

' -

Again H. polymorpha was cultivated in Reactor D in extended culture under identical operation conditions to those in the single stage tower, described in 7.2.3. In Reactor D perforated plates 0.5 mm in hole diameter were used as compartment separating trays. For the investigations in the absence of antifoam agents a 40 cm bubbling layer height was used which was controlled by the overflow. The basic problem in operating multistage bioreactors in the absence of antifoam agents had already appeared shorthly after the inoculation. The free volume between the bubbling layer and the tray above the layer were filled with foam. By microflotation effect

1'/L:200lh-' 1'/L=S0lh-'

f J // /j

/ //

A

//

Fig. 82. Cell mass concentration, X, as a function of the cultivation time, t, during the cultivation of H. polymorpha in Reactor D (three-stage tower) in the absence of antifoam agents employing substrate ethanol (for operation conditions see Fig. 81) 8 7 - 8 8 ) ; o 1. stage (at the top) • 2. stage

tfn«

0

i

8

12

16

h

24

a

3. stage (at the bottom)

Characterization and Performance of Single- and Multistage Tower Reactors

193

(see Chapter 4.2.8, 6.4 as well as 247>), the cells, which were enriched in the foam, passed through the tray into the next upper stage. Here they again were enriched in the foam and so on. Through this microflotation effect the cells were enriched in the upper stage; in the lower stages the cell concentration was considerably diminished as can be seen from Fig. 82. It was not possible to reduce this cell segregation by increasing the liquid circulation rate since the liquid transport capacity of the overflow was considerably reduced when foaming medium was present. The same flotation effect was also found by Kitai et al. 4 7 , 4 8 ) who also employed a (concurrent) multistage tower. To avoid this cell segregation they employed antifoam agent. From Fig. 82 the influence of this flotation effect can clearly be recognized. In the highest stage the cell concentration was highest and in the lowest one lowest. At t = 17.5 h the liquid recirculation rate had to be reduced, since the transport capacity of the overflow tubes were no longer sufficient to transport the foaming medium from the upper to the lower stages. This reduction of liquid circulation rate to 50 1 h _ 1 further reduced the cell concentration in the lower stage to very low values (0.5 g l" 1 ). Due to these different cell concentrations, the dissolved oxygen saturations were very different in the three stages: in the upper stage the lowest and in the lowest stage the highest dissolved oxygen saturations were attained. In the loop a high dissolved oxygen concentration could be preserved due to the high dissolved oxygen concentration and low cell mass concentration in the lowest stage, from which the medium was recycled. A comparison of X and 0 L / 0 * in the 1st (upper) and 2nd (middle) stages with those in the single stage tower (7.2.3) indicates that at the same X, the 0 L / 0 * in the multistage tower was higher than in the single stage one. Thus the OTR was higher in the multistage tower than in the single stage one. This is also shown by comparison of the volumetric transfer coefficients, k L a's, evaluated in single and multistage towers (Fig. 83). At long cultivation times the k L a-value in the three stage tower was about four times as high as that in the single stage tower. This yielded a higher maximum OTR (2.20 g 1 _ 1 h - 1 ) in the threestage tower than the one (1.28 g l" 1 h - 1 ) in the single stage tower. The considerable increase of k L a in the three-stage tower compared with the singlestage tower (Fig. 83) yields only a slight increase of the maximum cell productivity

Fig. 83. Comparison of the volumetric mass transfer coefficients, k L a's in Reactor C (single stage tower loop) and Reactor C and D (single- and three-stage tower loops) 8 7 ' 8 8 ) ; o single-stage • three-stage (h = 40 cm)

194

K. Schiigerl

from 0.71 (in single-stage) to 0.80 g 1 _ 1 h - 1 (in three-stage column). This is due to cell enrichment in the foam. Cell growth in the foam is possibly controlled by the substrate concentration and not by the dissolved oxygen concentration. To eliminate the foam two antifoam agents were employed alternatively: Desmophen and soy oil 8 7 , 8 8 ' 8 9 ) . The extended cultivation of H. polymorpha was carried out under the operation conditions specified in 7.2.3 and additionally employing an antifoam (Desmophen 3600) feed rate of 3 ml h ~ 1 . To test the influence of the bubbling layer height, h, on the OTR and cell productivity, Pr, h = 40 and 20 cm were employed in the threestage tower. The single-stage tower was operated with H = 160 cm. During the cultivations no foam formation was observed. Also no differences were found in the cell mass concentrations, X, of the three different stages. In Fig. 84 the volumetric mass transfer coefficients, k L a's, are plotted as functions of the cultivation time, t, for the single-stage as well as for the three-stage columns at h = 40 and 20 cm. One clearly recognizes that the lowest k L a values were found in the single-stage tower and the highest ones in the three-stage tower with h = 20 cm. The installation of trays improves the k L a values. This improvement increases with diminishing bubbling layer height, h. In all these systems, k L a first increases and then passes a maximum at X 3 g l - 1 (Fig. 84), similar to the k L a-behaviour found in concurrent tower loop reactors (7.2.1). This increase in k L a after inoculation is probably caused by surface active compounds secreted by the cells. The use of antifoam agents reduces k L a considerably as can be seen by comparing Fig. 83 with Fig. 84. This holds true for the single-stage as well as for the three-stage tower. With increasing cultivation time the oxygen uptake rate increases, the dissolved oxygen saturation diminishes and OTR increases. Again the highest OTR's were attained in the three-stage tower with h = 20 cm, and the lowest employing the singlestage tower, as expected. The same is true for the cell productivity: Pr = 0.69 g l - 1 h _ 1 (single-stage), Pr = 0.75 g l " 1 h " 1 (three-stage, h = 40 cm), Pr = 1.09 g l - 1 h " 1 (threestage, h = 20 cm). 400

0.100

-

f\ A

200 o -j Jc

/

A

ôS6

AS V—1

1

o.oso

O

—00

100'

0.025

Fig. 84. Volumetric mass transfer coefficients, k L a's, as functions of the cultivation time, t, during the cultivation of H. polymorpha in Reactors C and D employing substrate ethanol and Desmophen 3600 as antifoam agent. (For operation conditions see Fig. 81) 87, 88).

o single-stage • three-stage, h = 40 cm A three-stage, h = 20 cm 8

12

16

24

Characterization and Performance of Single- and Multistage Tower Reactors

195

The operation conditions were again identical with those given in 7.2.3, when soy oil with a feed rate of 0.5 ml h - 1 was employed as antifoam agent 87'88>89>. The soy oil was a less effective antifoam agent than Desmophen 3600. When employing soy oil the foam did not disappear, only its structure changed: the bubble size increased and the foam became less stable. After 8 h of cultivation, the cells were enriched again in the upper stage due to their microflotation (see Chapters 4.2.8 and 6.4 as well as 2 4 7 ) . However, the nonuniformity of the cell concentration was considerably less than for the antifoam-free system, the cell concentration in the upper stage was, at its maximum, by a factor of two higher than the one in the lowest stage. To reduce the foam amount at t = 8 h the antifoam feed rate was considerably increased for a short period of time and then decreased as can be seen from the upper part of Fig. 85. In the lower part of this figure the cell mass concentration, X, OTR, and the ethanol uptake rate are plotted as a function of the cultivation time, t. After inoculation the cells began to grow without a lag phase. In this exponential growth phase IV = 0.11 h _ 1 prevailed. During the time t = 6 to 9 h a reduction of |xm occurred. After this period a new exponential growth phase began with |i m = 0.26 h - 1 . Obviously the cells adapted themselves to the soy oil and used it as a substrate.


Operation

Batch, VR = 3401h" 1

Continuous, VR = 14001b" 1

Stages

Single

Ten

Single"

Ten"

Gas distribution

Porous plate

Perforated plate trays

Porous plate

Perforated plate trays

H™ ( h - ) Yx,sc ( - ) RQ ( - ) Pr, Pr* (g 1 _ 1 h" 1 )

0.54 0.80 0.88 1.63

0.52 0.75 0.93 1.66

0.36-0.44 1.16 0.93 2.64

0.36 0.53 0.86 1.15

" Pr = DX in continuous culture;

b c

Pr = — - in batch culture (not considering the lag phase); At H = D formal value calculated with regard to glucose

Table 11. Comparison of batch and continuous cultivations in stirred tank reactor by employing E. coli and antifoam agent (Desmophen) 3051 Cultivation

Batch

Continuous

^(h"1) Yx/s(-> RQ ( - ) Pr, Pr*

0.90 0.70 0.87 1.25

1.10 0.90 0.85 2.0

" Pr = DX in continuous; Pr = —— (lag phase is excluded)

measurements, that the stage separating trays strongly influence k L a, as can be seen from Fig. 88. Thus the gas dispersion efficiency of the trays is an important factor for multistage tower reactors. One can recognize from Table 12 that the three-stage tower loop (Reactor D) has a higher performance than the single stage one (Reactor C), if antifoam agents are employed. (In the absence of antifoam agents cell microflotation deteriorates the performance of Reactor D, with regard to cell productivity.) One can also recognize from Table 12 that the height of the bubbling layer also influences the multistage tower loop reactor performance. This influence has been well represented by means of model media (Fig. 89). As expected k L a, OTR and Pr increase with diminishing height of the bubbling layer, h. This phenomenon is due to the higher kL-value which prevails immediately after bubble formation and the smaller d s caused by the slight bubble coalescence in this short layer. The liquid recirculation ratio, influences the performance of tower loop reactors in different ways. Increasing y reduces E G , a and k L a, if OTR is dominating (Fig. 64). It has an optimum value, if substrate limitation prevails (Chapter 7.1.1/7.1.2) and it can

202

K. Schügeri ) ¿1.2 • 1.66

1.0

0.6

/ L

0.4

J Vj

0.2

—

Wsg

-

Fig. 88. Volumetric mass transfer coefficient, k L a, in a six-stage tower reactor 20 cm in diameter, as a function of the superficial gas velocity, wSG for different trays. 1 % ethanol and nutrient salt solutions 87) perforated plate hole diameter, d H , number of holes, N H , relative free cross section Q H d H (mm)

Ng(-)

QH(%)

1 % ethanol

nutrient salt sol.

3 1 0.5

214 248 558

3.82 0.62 0.35

&

A

• 0

• •

influence the longitudinal medium dispersion, especially the longitudinal cell mass concentration profiles. The latter effect is especially important, if cell sedimentation due to cell flocculation is considerable. In this case, with increasing y, the tower loop reactor performance improves (Table 13). With decreasing y, substrate conversion, U s , and yield coefficient, Y x/ s diminish in single-stage tower loop reactors. Also the cell productivity, Pr = DX, diminishes from 2.64 g 1~1 h~ 1 (y = 70) to 1.48 g 1~1 h ~1 (y = 0). The same holds true for ten-stage tower loop reactors (Table 13). The cell productivity, Pr = DX, also diminishes from 1.15 g 1 _ 1 h ' 1 (y = 13.8) to 1.08 g 1 _ 1 h - 1 (y = 0.95) if the recirculation ratio, y, is reduced. 7.3.2 Influence of Medium Composition The influence of substrate type on the biological parameters is well documented (Tables 4 and 5). In the present case (xm is much higher for glucose than for ethanol or methanol. Also the substrate and oxygen yield coefficients as well as

Characterization and Performance of Single- and Multistage Tower Reactors

•C H

r» ts o 5r o Vl o o (N Ö •o

0.18

•fi

§

Ö

—

o on

S O. ;

•

S

Qo,

• 0 L /0f o X

The addition of antifoam agent (Desmophen) to the system reduced n m (e.g. in Table 12 from 0.25 to 0.21 h - 1 ) . It also diminishes k L a and OTR, but increases Y x / 0 (see single-stage in the absence of antifoam agents and in tha presence of Desmophen in Table 12). The strong reduction of k L a due to antifoam agents is also well documented in Ref. 218) .

Characterization and Performance of Single- and Multistage Tower Reactors

209

Fig. 95. Cultivation of H. polymorpha in Reactor A employing substrate glucose. Influence of the short time lack of substrate. Aeration rate: 0.55 vvm 1 6 , 1 5 5 ) . (For symbols see Fig. 93)

M

aera t ioi off/on

3 D)" \

1

ir n MODEL

reference inputs OPTIMIZATION • —

adaptive control

0= unit

_J

Fig. 101. Open loop feedback optimal control (OLFO) 3 0 9 ' 3 1 0 )

also their adaptation to their environment under different operation conditions (substrate, oxygen transfer limited growth ranges) and such phenomena as microflotation as well as cell sedimentation and flocculation. It is expected that by means of this additional information and improved models it will be possible to simulate the complete process in pilot and production plants based on the bench scale reactor measurements and to control it by one of the advanced techniques. A good beginning was made by Luttmann 181) , who simulated an operating pilot plant reactor by means of the bench scale masurements of H. Buchholz 16) and Lehmann et al. 2 6 ) as well as by the information provided by the pilo-plant operating company, assuming different economical conditions. Further developments along this line are necessary to realize these goals.

10 Acknowledgement The author acknowledges the financial support of the Ministry of Research and Technology of the Federal Republic of Germany, Bonn, and the German Research Foundation (DFG), Bonn, and thanks Dr. I. Adler, Dr. H. Buchholz, Dr. R. Buchholz, Dr. R. Luttmann, Dr. J. Voigt, Dr. W. Zakrzewski and Mr. J. Lippert for their excellent cooperation as well as Prof. Dr. H. Sahm, Institute of Biotechnology, KFA Jülich, Prof. Dr. M. Thoma and Dr. A. Munack, Institute for Control and Automation of the University of Hannover for their support.

214

K. Schiigerl

11 Nomenclature L (length), M (mass), T (time), K (temperature), — (dimensionless) area gas/liquid interfacial area geometrical gas/liquid interfacial area specific interfacial area a calculated by Eq. (35)

L2 L2 L2 L~ L"

specific interfacial area

L~

Bo = ^ ^

Bodenstein-number in the tower

—

BSA

Bovine Serum Albumine

A A a' a = A/V l a A

V DF

S C, = — S0

dimensionless substrate concentration at the reactor exit —

X C x = — ——— dimensionless cell mass concentration X + Y X/sS0 c concentration D dilution rate Da = |imT, Damkohler number DM diffusivity Df longitudinal liquid dispersion coefficient in tower Dg longitudinal gas dispersion coefficient in tower dB mean bubble diameter Eq. (32) ds Sauter bubble diameter Eq. (33) de dynamical equilibrium bubble diameter dp primary bubble diameter Eo2 aeration efficiency Eg mean relative gas hold-up, Eq. (31) EF mean relative liquid hold-up E energy dissipation rate (local power input) Ej(n) one dimensional frequency power spectrum E t (k) one dimensional wave number power spectrum F liquid feed rate g acceleration of gravity H height of the aerated layer He Henry coefficient Hl height of the bubble free liquid layer h height of the aerated layer in multistage tower I ion strength Eq. (6) I = u'/0 relative turbulence intensity K = K0 Ks Kst

s — dimensionless substrate saturation constant So + Xo/Yjys oxygen saturation constant substrate saturation constant coalescence constant, Eq. (97)

— ML-3 T"1 — L2T_1 L2T-1 L2T_1 L L L L T2L-2 — — ML2T-3 L 3 T~ 3 L2T-1 L3T_1 LT~ 2 L T~ 2 L 2 L L ML-3 — — ML-3 ML-3 —

215

Characterization and Performance of Single- and Multistage Tower Reactors

k = 2jtn/0 kL kLa kLaE kLaa kLa„ kc

wave number mass transfer coefficient volumetric mass transfer coefficient k L a at z = o Eq. (97 a) k L a" in range 0.1 ^ z g 1 Eq. (97c) kLa" at S = 0.0 coalescence rate constant Eq. (116) relative mole mass of i Mi M coalescence factor Eq. (100) L m = k L a/(k L a) H2 o coalescence factor Eq. (18) m* = (k L a) corr /(k L a) ref coalescence factor Eq. (19) n frequency O dissolved oxygen concentration in medium O in loop OB O in tower OF O in liquid oL Oc critical O in liquid O* O at saturation O f at saturation Of OTR oxygen transfer rate oxygen uptake rate OUR P gas pressure integral power input PL Pr cell productivity tower cross section Q loop cross section QB cross correlation function R(RX) Rc coalescence rate Eq. (116) gas constant Ro Rx =

R* ——- dimensionless cell growth rate

X 0 + Y X /sS 0

Ro ROB ROF

Rs RSB RSF R Ì RAB RE

RQ r

x

S S* SB S

F

=

£/£r

oxygen uptake rate oxygen uptake rate in loop oxygen uptake rate in tower substrate uptake rate substrate uptake rate in loop substrate uptake rate in tower cell growth rate Eq. (1) cross correlation function Eq. (47) autocorrelation function Eq. (48) respiratory quotient point distance in x direction substrate concentration in medium foam stability S in loop S in tower

L"1 LT" T

- I

T

- I

T

- I

T

- I

T

- I

—

— —

—

- I

T

ML ML ML ML ML ML ML ML ML ML ML : ML

3-p-l

V

L2 —

LTL2T —

ML - 3 - p - l ML - 3 - p - l ML - 3 - p - l ML - 3 - p - l ML - 3 - p - l ML - 3 - p - l ML - 3 - r - i — —

—

L ML - 3 -

ML - 3 ML - 3

K. Schiigerl

216 S in feed S„ S at which k L a" = 2kLag SD T temperature turbidity temperature, Eq. (20) Tt t time u velocity tJ local mean liquid velocity bubble rise velocity UB u s == i - c . substrate conversion u velocity measured by anemometer U„U y turbulence velocity fluctuations u' turbulence intensity Eq. (45) U G == w s g / E g gas velocity Vr UB = liquid velocity in loop Qb UF == w SL /(l -- E g ) liquid velocity in tower V volume of aerated layer in tower volume of bubble free layer in tower vL volumetric gas flow rate vG = V + F volumetric liquid flow rate in tower vL = R volumetric recycle rate vR bubble swarm velocity, Eq. (34) wBs W relative bubble swarm velocity with regard to liquid, BR Eq. (34a) vG superficial gas velocity in tower WSG Q WSL

vL Q

X XB xF x„ X X* XoG x co 2 Xqg Yx/o Yx/s Y z = x/H y = Af

Vr VL

ML"3 ML-3 K K T LT" 1 LT" 1 LT" 1 —

LT" 1 LT-1 LT-1 LT-1 LT" 1 LT-1 L3 L3 L3T-1 L3T-1 L3T-1 LT" 1 LT-1 LT-1

superficial liquid velocity in tower

LT" 1

cell mass concentration X in loop X in tower X in feed longitudinal coordinate in tower (x = o at aerator) longitudinal coordinate in loop mole fraction of 0 2 in gas mole fraction of C 0 2 in gas Xqg at the inlet x = o oxygen yield coefficient substrate yield coefficient mole fraction dimensionless longitudinal coordinate in tower

ML"3 ML"3 ML-3 ML"3 L L — — —

— —

—

medium recycle ratio

—

dynamic viscosity of liquid macro scale of turbulence, Eq. (52)

ML-1' L

217

Characterization and Performance of Single- and Multistage Tower Reactors

Xf Hm H v gL eG E £r a T xE